What does it mean?

How does the IT scene change with the implementation of hybrid cloud computing? First let’s consider private clouds only. In the following, I will use an enterprise data center and its private cloud interchangeably for the ease of discussion, although not all data centers have been converted to private clouds yet. Some company may have several data centers (and therefore private clouds) in the US, or even worldwide, across multiple time zones. So even before talking about hybrid, using this technology we can combine those physical data centers into one single logical private cloud. A logical cloud consists of physical private clouds (data centers) and may be recognized as one entity.

Logical private cloud

With a logical private cloud, using some technologies from CloudVelocity, we can move applications that may consist of physical machines (PMs: not virtualized) and virtual machines (VMs) anywhere and anytime we choose. In the following figure, we can pass PMs and VMs back and forth seamlessly between our home cloud and any other private clouds of our company. Although it shows only a subset of interactions below, we can potentially move PMs and VMs in any way that makes sense by some predetermined criteria. It may that one PM or VM is passed to another cloud and then to the third one and so on. It would become pretty complex to manage your Pms and VMs under such a new paradigm.

PMs and VMs move around only among private clouds owned by the same organization. A set of such private clouds may be considered as one logical private cloud.

This means we can finally implement several things discussed in part 1, including:

• Follow the sun
• Follow the moon
• Load-balancing
• Fail-over/disaster recovery

Follow the sun

In a given workday, access to software applications and utilities running on servers and other IT equipment—and therefore clouds—fluctuates. Access starts to grow as people start their day’s activities in the morning, it hits a peak, then subsides towards the evening. Access is lowest during the night. So you might want to move your PMs and VMs to other time zones where the sun still shines and more loads need to be processed. We can expect a better response time when loads and processing units are close to each other.

Follow the moon

In many countries, power is cheaper during off-hours (normally nights, hence follow the moon). Sending your Pms and VMs to such time zones may reduce your operation cost. Additionally, even within the US, power cost can fluctuate hourly if a variable power pricing model is implemented and applied to data centers. By shifting your VMs to a data center whose region gets the lowest power cost, you may save on running costs.

Load-balancing

Just as we load-balance among servers at a data center, we may want to send loads to several different private clouds. In this way, when one data center gets very busy, such loads can be passed to other data centers to share the burden. How you move PMs and VMs should be determined by predefined metrics to optimize your operations for a few factors, such as operating cost, response time, and throughputs. Each organization has its own goal for its operation, and the metrics should be tailored to accommodate it.

Cloud bursting may be related to load-balancing, although it is not the same. When a load increases in a private cloud, we may want to move all or part of it to a public cloud for on-demand processing; this is known as cloud bursting. PMs and VMs that are processing the load can be moved to a public cloud for continuous processing. When the load subsides, PMs and VMs on the public cloud can be disabled. There has been a lot of talk about cloud bursting, but now it can become a reality. We need a good automated system to move PMs and VMs, and to enable and disable them as needed. A good policy is a must-have for this as well.

Fail-over/disaster recovery

The San Francisco Bay Area will have a major earthquake some day for sure, and when it happens, much of the existing infrastructure, including data centers, will be unusable. If we have a way of duplicating what we are running in our primary data centers at a secondary site far enough away (such as the Sacramento area, a little more than 80 miles from the Bay Area) and transferring execution state information intact to the distant site, processing could proceed without interruption.

Super logical private cloud

With this technology, we do not have to consider the boundary between private and public clouds either. So the logical private cloud can include public clouds, becoming a super logical private cloud, or what I call a supercloud.

A green oval depicts a private cloud, and a light-blue one represents a public cloud.

This configuration would make managing PMs, VMs and clouds much more complex. We can move our PMs and VMs between private clouds, between private and public clouds, and among public clouds. We will no longer be restricted to a move between one cloud and another cloud (a one-to-one move) but can implement one-to-many and many-to-many as well. Then it will become necessary to develop a system that allows automation. As we involve many private and public clouds of various implementations, we will not be able to easily track how to optimize such moves. For that, we will probably need a policy based on predefined metrics. Cost may be the number one factor. But at the same time, we want to maximize response time—and the performance of developers scattered around the globe.

Also, note that many superclouds may share the same private and public clouds. This means that loads at each private and public cloud could fluctuate over time. So depending upon how busy each cloud is, we may want to dynamically alter how we form a super logical private cloud for optimization.

By the way, when a supercloud is developed and deployed, will we call it a supercloud or simply a cloud? Those IT folks who will follow us in the future may take it for granted and consider it a normal IT deployment and execution environment. Throughout IT history, when some technology or method becomes transparent as part of an overall system, that is when we say that that technology really has matured.

Who uses hybrid clouds and benefits from them?

I can think of three parties, although there may be more.

Enterprise end-users

Enterprises that have their own private clouds can extend them to public clouds to produce hybrid clouds to exploit the things I mentioned above.

Data center providers

If you are a colo provider, you can sell extra services at your center to realize hybrid clouds for your clients. There are different levels of providers. Some may simply rent a space, while others provide both equipment and services. Some may provide both private and public clouds at the same data center. For them, this is a perfect tool to increase their revenue.

Third-party consultants/service companies

If a colo provider does not want to provide any service other than space, those guys with the hybrid cloud technology can help end-users implement hybrid clouds.

Energy consideration

Finally, my blog always ends with a question about what the subject means to energy efficiency. Although inconclusive, there has been some discussion about whether cloud computing is more energy efficient than its predecessors. I think it depends upon whose view you take. If you are a user, you pass some or all of your computing needs, along with support staff, software, hardware, power, cooling, water, and other things, to your cloud provider on an on-demand basis. Since you can reduce your investment on these, it is certainly energy efficient for you. It may or may not be for your provider. If the provider has very little utilization of their facilities, they may not be profitable or energy efficient at all. You may still have to have a large staff, a large space, dedicated IT and facilities equipment, facilities support such as cooling, and so on. That cannot be very energy efficient.

When a hybrid cloud becomes a supercloud and our energy becomes more scarce, we may need to look at energy consumption and energy efficiency at the supercloud level without distinguishing private or public clouds, which may sound silly at this point. It is because the US seems to be doing fine for the foreseeable future with shale gas and oil, but who knows what may happen next?

James Urquhart

There have been a lot of talks on cloud computing, about what it is, what kinds exist, and how it changes the way IT is delivered. Virtualization, security, automation, provisioning, tools, and other details have been discussed many times by many people. But his talk was the first I’ve heard that specifically covered how to design the entire system of cloud as a whole.

If I may oversimplify his point:

• Cloud is an interconnected complex adaptive system.
• The interconnected system as a whole may collapse, even if we carefully design and operate each component of the system.
• It is vital to build resilience into the design and operation of such a system.

He gave an example of a market crash, known as Flash Crash, which occurred on March 6, 2010. I myself did not remember it, and neither did most of the audience. An automated system for stock exchange executed the sell program extremely rapidly in just 20 minutes. With this, the market got more active and more orders poured in. The system spun more threads to process these orders, and finally it got out of hand and crashed. With this story, James introduced the term complex adaptive system.

It is defined as:

Further,

where emergent behavior or emergence is defined as:

the way complex systems and patterns arise out of a multiplicity of relatively simple interactions.

Because each component of the complex system is relatively simple and easy to control, we tend to think we can control the system as a whole. But as the discussion above shows, that is not the case. If there is no way to predict future behavior, the only thing to do is to accept potential failure and get ready for it. Because there is no way of knowing how the system may fail, we may randomly produce a large number of problems to test our systems. James showed one such system that causes random chaos, Netflix’s chaos monkey.

I usually relate the subject of this blog to green IT or energy. That’s hard to do in this case. One thing I can say is this: if the complex adaptive system fails and starts to behave erratically, that surely wastes energy, because the behavior does not produce any meaningful outcome. But I need to think further about what this means to green IT.

If you want to know more about it, you can read some of his blogs.

• Exploring cloud and complexsystems

• Five resources for learning complex adaptivesystems

• The art of failure in cloud’s complexsystem

James recommended reading Drift into Failure by Sidney Dekker of Griffith University, Australia, and other books, shown below, to further understand this.

In the opening speech of the 5th SVLG Data Center Energy Summit, Anne Smart, Director of Energy, Silicon Valley Leadership Group (SVLG), gave a good introduction to this topic as well as to Andrew Feldman, Corporate VP and General Manager, AMD, who delivered the keynote speech. You can watch their speeches here (a little over 16 minutes).

Andrew was CEO of SeaMicro, which developed and marketed low-power servers and was acquired by AMD earlier this year. Because the video is less than 20 minutes, you may want to invest your time in watching it yourself. But his point was simple. There is an area where we made significant progress in energy efficiency for data centers, but there is another area where we did not make much progress. The former is the facility side and the latter is the IT side. A lot has been done about energy efficiency in the facility area, such as finding better ways of cooling (hot/cold aisle and air economizer), which Andrew called low-hanging fruit. Now cooling efficiency has improved to three one-hundredths of the power required to run a server.

Now about IT. Let me inject my thoughts here before going on with Andrew’s speech. Before he started discussing the energy efficiency of IT in the data center, I knew I was going to agree with him. We have been trying to alleviate the symptom of data center energy problems by attempting to control cooling and power rather than curing the root cause of the data center energy crisis, that is, efficient ICT equipment and making better use of it, that is, running it without utilizing its full capacity. Emerson in its energy logic claims that IT is the root cause of the large consumption of power in data centers and we had better control IT. We are beginning to pay more attention to IT as we try to control data center power usage. We need to pay more attention to using more energy efficient equipment, such as SeaMicro’s, and using virtualization to increase the utilization of each server. We need more than PUE to gauge and measure IT efficiency. There have been a few data center metrics proposed, such as McKinsey’s CADE. The Japanese have proposed DPPE. Both metrics take IT’s behaviors into consideration. I recently found an interesting metric for IT energy efficiency called Par 4. I have not reviewed it yet, but the short description I read sounds interesting. I plan to write a separate blog on it.

Andrew started to talk about IT (servers) in terms of data center energy efficiency. He dealt with server energy efficiency at SeaMicro with low-power chips before he came to AMD. And he is one of the most qualified persons to discuss the subject. He bluntly stated that he was disappointed at the progress we have made in that department, because a server is the most powerful consumer of power in a data center and needs more attention.

He mentioned that like anything else energy efficiency improvement for servers began with the low-hanging fruit. In this case, it was the power supply. The power supply used to lose 18% in transit from PDU to server. Now with advances in power supply technology, it is down to 7%. But the real meat in energy efficiency is the server itself. At SeaMicro, Andrew and his team developed a server that requires only a quarter of the power necessary for other servers. But it took five years and $50M, which was not readily available from the VC community in the funding climate then. He is thankful to the Department of Energy and the State of California for their grants that made it possible for SeaMicro to complete their servers.

He wrapped up his talk by saying that we should not stop innovating in energy efficiency in both facility and server technologies. Power is such a big problem in the data center that we need to keep working on it.

Scott Jarr

When I was researching the NoSQL segment, I found it confusing enough, but there is also a NewSQL movement, which confused me further. The NoSQL movement began in an effort to accommodate the Big Data phenomenon. In the traditional database segment, ACID—atomic, consistent, isolated, and durable—is of utmost importance. The relational database was developed to guarantee ACID and for transaction-oriented applications. The traditional relational or SQL database is fine, as long as the data comes in at a reasonable speed and volume and is of limited variety. But at some point, these parameters exceeded what the traditional SQL database could handle, and new ways to cope with them were increasingly required. That is where NoSQL comes in. NoSQL, in general, does relax some of the rigid SQL rules (abandoning SQL partially or altogether and thus ACID) and accommodate these new requirements; i.e., scale-out, high availability (HA), replication, and performance. Therefore, NoSQL in general does not have SQL, relational schema, joins, or ACID (this is obvious since these are traits of the relational/SQL database). Scott put the comparison of Old SQL, NoSQL, and NewSQL on a piece of paper as we spoke. I reproduced it here. Old SQL (yet another term) refers to the traditional relational/SQL database that dominates the enterprise world.

Comparison of Old SQL and NoSQL

So in other words, in order to gain scale-out, HA, replication, and performance, NoSQL abandoned SQL/relational schema partially or altogether. What NewSQL is saying is that it can accomplish every feature in the table above while keeping the relational/SQL schema (and therefore ACID and join).

If that table is expanded with NewSQL, we have the following.

Comparison of Old SQL, NoSQL, and NewSQL

How can that be possible with NewSQL? Performance gain is a result of new architectures that remove the old baggage of OldSQL, many leverage memory for additional improvements. Actually, Gigaspaces, which I also interviewed and will write a blog about later, has an in-memory cache technology working with other NoSQL companies. However, Michael Stonebraker, CTO of VoltDB, said in one of his talks (available in a 30-minute video) that running in memory alone does not guarantee the performance gain needed to accommodate the speed at which Big Data comes in.

Mike explained in his talk that there is nothing wrong with the concept of SQL itself. It is the implementation of SQL that causes the problems shown in the table. Because of the less than perfect implementation, 96% of the time is spent on overhead and only 4% on useful work, as indicated in the following graph extracted from his presentation.

CPU cycle use in a typical SQL implementation. Most of it—96%—is used for overhead.

Unless these overheads are removed, even if all data is placed into memory, extreme performance improvement is not expected, because it only addresses the 4% but not the rest of the 96%. Typical NoSQL databases abandoned or partially supported SQL to bypass this problem. VoltDB faced the current inefficient implementation of SQL and developed their version of the SQL database from the ground up to eliminate these overheads. I am not covering each overhead in detail, but you can watch his easy-to-follow video.

OK, I get it. Then, what does this mean to the whole area of NoSQL? Does this mean the whole area of NoSQL gets consolidated into a single technology like NewSQL? Scott drew me a good figure to explain this, which he had already published in his own blog posts (the figure below came from part-1 and part-2).

From Scott Jarr’s blog. On the Y axis, data speed, size, and complexity grow upwards.

There are five areas to address in the enterprise in terms of data collection and analysis (analytics): interactive, real-time analytics, record lookup, historical analytics, and exploratory analytics.

The five areas are further explained in the following figure with applications and time scale.

Note that VoltDB is colored differently from the rest of NewSQL in the graph, but that is meant to emphasize its position in the NewSQL group. VoltDB falls into the NewSQL camp. Scott emphasized its performance superiority over others. The performance benchmark they share is 3 million transactions per second (TPS). According to Scott, the traditional RDBMS is trying to cope with the Big Data problem (velocity, volume, and variety) by scaling up (throwing in more CPU and storage power rather than using parallel computing).

In Stonebraker’s video, he said that VoltDB was five times faster than Cassandra and also faster (he did not say how much) than an unnamed incumbent’s database. When I consulted with MySQL, it was very fast (before their 5.0, which incorporated enterprise-ready features) and faster than this incumbent’s, but they could not publish the benchmark for fear of a lawsuit. I can understand that. When I consulted for JBoss, a Japanese open-source consortium compared their performance with other products like IBM’s Websphere, without any tuning. The number was not very good, mainly because those who ran the benchmarks did not know how to tune JBoss’s compared to IBM’s. After a JBoss engineer flew over there and tuned it, it improved drastically. So when we conduct a performance comparison, we need to set up a ground rule for comparison for every participant.

Of course, there are some overlaps among those technologies and their areas of applications, but this figure is a good picture of how each technology is suited for its application area. Hadoop is batch processing based and is not suitable for real-time analysis. Many people think Big Data and Hadoop are synonymous, but this clearly shows they are related but not the same. In the utilities business, a large amount of meter-read data gets collected, aggregated, and stored. By daily or monthly analysis of power usage for a particular area, a utilities company can probe into usage patterns and trends. Actually, some utilities are using Hadoop now, according to the Soft Grid conference.

Scott thinks NewSQL, NoSQL, DataWarehouse, and Hadoop will remain separate technologies because each of them is suited for some specific area of data collection and analytics. But he advocated that these areas and their tools be tightly integrated to provide analytics and thus effective real-time actions, as in the following figure. By incorporating the analytics results for long time spans into short-time analytics, more effective actions could be obtained.

Finally, he showed the current applications of VoltDB, as follows.

Google has been in the PaaS space but recently dived into IaaS. A great thing about living in Silicon Valley is that local meetings with the movers and shakers of technology market segments are within an easy 30-minute drive. The meeting that prompted this blog was on cloud computing. Cloud computing is changing the way IT is delivered and thereby changing how and where we conduct computing. Consequently, it has a vast impact on the way we use energy. Some conclude cloud computing saves energy, while others are skeptical about it. In any event, Google, a mammoth of computing power, has entered another category of cloud computing, so I attended SVforum’s cloud computing and virtualization special interest group meeting on Google’s new service. (By the way, Google Compute Engine was announced about a month ago at the Google IO conference.) The SIG meeting was well attended; more than 100 people showed up.

These were the speakers:

• Marc Cohen and Kathryn Hurley of Google, developer relation engineers
• M. C. Srivas of MapR, cofounder and CTO

Dave Nielsen, one of the cochairs of this SIG, and others did a good job organizing this meeting. Normally, when a new product or service is introduced, a vendor or provider gives a dry presentation about how great their product or service is. Google wanted to show how solid their high speed and security offering is, so they included a couple of good demos.

Well, that was not enough. On top of that, they invited MapR, one of their partners, to share their experience in deploying MapR’s version of Hadoop on the Google Compute Engine platform. Hadoop is open source from the Apache foundation and is used to crunch Big Data. MapR took Hadoop and added features to make it compatible with enterprise requirements.

Big Data and Hadoop have opened up a tremendous number of market opportunities, so MapR is not the only one to exploit them; Cloudera and Hortonnetworks also provide Hadoop on steroid versions for enterprises. By the way, I found a research report by Dave Ohara, The big machine: creating value out of machine-driven bigdata, to be very good tutorial material on Big Data.

MapR also has deployed its engine on Amazon AWS, and a performance comparison would be interesting. During the discussion, MapR revealed their performance data but said they had not conducted such a performance benchmark on AWS.

Back to Google Compute Engine. Google is no stranger to cloud computing. Here’s a simplified list of Google’s cloud offerings:

Marc Cohen speaks about Google’s history of cloud offerings.

What is Google Compute Engine? Marc summarized it on one slide:

Marc Cohen presented an overview of Google Compute Engine.

In short:

• Infrastructure as a service (IaaS)
• Supports Ubuntu and CentOS (more operating systems, such as Windows, will come later)
• KVM as hypervisor
• Deployable in two territories (one eastern and two central time zone data centers, only in the same data center rather than inter–data centers)
• In private beta (need to apply to be deployed on the platform; see here for more detail)
• Free for now (only qualified users) but will be charged for later
• No SLA guarantee now (under consideration when released officially)

Marc did not elaborate on SLA, but judging from what he said, you need to specify which territory you want to deploy your load in. All the computing and data associated with it stays in the same data center (i.e., cloud). No matter how we improve technologies, we will still be bounded by the laws of physics, and we cannot send packets any faster than the speed of light. If they want to guarantee SLA, they need to make a lot of assumptions and impose restrictions on their customers. I have not heard any cloud service provider discuss SLA, and I wonder how they can guarantee it, even with conditions.

More details can be found in their data sheet here, and here are two useful links:

• Google’s Cloud platform
• Google Compute Enginer Documentation

Some architectural information follows.

The yellow disklike box behind Kathryn Hurley is cloud storage.

One thing unique about this is the use of their command language interface with gcutil library.

Now to benchmarks and usability shared by MapR. It is more convincing if your actual user, rather than yourself, says good things about your offering. Even though this is not a blog on Hadoop or MapR, here’s some basic information shared by Jack Norris, VP marketing, of MapR.

MapR’s summary

He also summarized MapR’s deployment on Google Compute Engine, as follows:

Srivas added the following, more-detailed benchmark:

The one on the Google platform outperformed that hosted on a physical platform in every benchmark except for processing time:

There was no comparison in cost. Srivas said jokingly that we had better think twice before purchasing and owning a bunch of servers. This is because a server’s life is probably only two to three years, and the minute you buy a new server, it becomes obsolete because new servers with new technologies are invented constantly. Spending so much money is one thing, but after your business goal changes and you no longer need that many servers, what do you do with them? They do not disappear magically.

In the area of IaaS, AWS was way ahead of the market curve, followed by RackSpace and others. As Dave Nielson said at the beginning of the meeting, those who were working on IaaS (e.g., Amazon) are adding a PaaS solution (e.g., Elastic Beanstalk) and those who were in the PaaS market (e.g., Google) are adding an IaaS solution. The cloud market is still expanding, and in spite of some problems, such as lock-in (due to no standards), security worries, and lack of control, because of its wide and broad market, it is still growing rapidly. As long as mobile computing and sensor networks (such as smart grid) are growing, it does not seem that the end of growth is even near. When very few interoperability cloud platforms exist, it is always good to have competition.

Because a lot is going on with IT, this panel was very interesting. As more and more people and devices and equipment (without human intervention) get online, the importance of data centers increases even more.

First, here is a list of panel participants:

Moderator

• Bruce Myatt, M+W Group

Bruce Myatt

Panelists

• Peter Gross, Bloom Energy
• Ron Hughes, State of California
• Zahl Limbuwala, BCS Chartered Institute of IT
• Ben Radhakrishnan, Lawrence Berkeley National Laboratory

From left: Peter Gross, Ron Hughes, Zahl Limbuwala, and Ben Radhakrishnan

The following is a summary of the discussion, with my comments (indicated by ZK). Bruce started by saying how, since the 2007 EPA report on data center energy consumption, we came to increase data center energy efficiency by means of several technologies and methods. Those include air economizer (free cooling), cold and hot aisle separation, cold or hot aisle containment, and measuring, which led to the wide deployment of the PUE metric and more energy efficient IT equipment. Other than more energy efficient IT equipment, these are primary from the data center operator’s view. Then, Bruce posed two questions.

Question 1: Types of changes in IT operations in the next 5 to 10 years

The first question was the type of changes taking place now and what will happen in 5 to 10 years that will have a big impact on IT operations.

A number of such changes were discussed, such as:

1. Emergence of mobile devices (specifically tablets, whose shipment will reach 400 million in 2016)
2. Cloud computing/virtualization
3. More security needs
4. Much bigger scaling changes for each server refresh
5. Emergence of ultra–energy efficient HPC gear for supercomputers that encourages cloud computing operations
6. Change of emphasis from facilities to IT
7. The CIO’s role will change to that of a broker of services who decides where each application resides, either in-house or hosted, from that of a provider of infrastructure services.

(ZK: Most points raised were not accompanied by a detailed explanation, and I would like to express my own opinion about their impacts on data centers. The second question below was supposed to answer that. However, there was no clear one-to-one correspondence between the questions in question 1 and the answers given in the discussion of the second question. That does not mean the discussion was not useful, though. For reference purposes, I’ll refer to each question as 1-x, with x being the number associated with it.

Actually, 1-1 through 1-5 are very much related. Although virtualization is not cloud computing, it constitutes an important component of it. In short, virtualization with factors like policy and automation would create a cloud. As we have more mobile devices like smartphones and tablets, it becomes necessary to take care of them at a data center. Because the scaling factor would be several orders of magnitude bigger, because of more new IT equipment processing power and the sheer number of client devices, it will become necessary to have an infrastructure that can handle such a scale. That is cloud computing. As computing is concentrated in clouds, more security needs will be emphasized. Security is not limited to perimeter security of a data center. Security can be broken within the data center and via networking as well. Hacking and malware will become more sophisticated, and when security is bleached, its consequence would be several orders of magnitude greater than when everything was processed manually on paper.

As to 1-6, I also felt that more emphasis was placed on the facilities side in data center conferences before. But conferences, media, and blogging are now beginning to pay more attention to IT for improved data center operations and energy efficiency. This is very important because, after all, facilities exist to support the reliable and stable operations of ICT equipment. In turn, ICT equipment is run to support the business goals of an enterprise. It is easier to see concrete numbers like kilowatt hours consumed and/or temperature and humidity than to see what applications are doing for services.

As for 1-7, cloud computing with automation will relieve a CIO and his staff from repetitious and routine tasks, and they will become knowledge workers. See my previous blogs on this subject here and here.

As cloud computing moves into the mainstream of computing, some operators will abandon their own data centers and move their computing to cloud providers. Although data centers will not disappear, the number of data centers will decrease. At the same time, cloud providers (cloud data center operators) will expand their current data centers at their current and new locations. Each cloud data center in the future may be bigger than current ones, requiring more power and cooling. Or future data centers may require less space and power (thus, less cooling energy) because of new technologies like supercomputers with an enormous amount of processing power and very high energy efficiency.

Question 2: How those changes with IT will affect the design and operation of data centers

The second question was to solicit opinions on what those changes would bring to data centers.

(ZK: As I said, there was very little one-to-one correspondence between the first question and the second question. That is why I injected my own opinion above.) These were some of the points raised during discussion of the second question:

1. A method of assessing data centers from environmental, economic, and specifically benefit points of view will be developed.
2. One proprietary system providing IT functions (hardware, software, server, storage, and network in a bundle) will be used more.
3. Self-management, including self-healing, in the IT segment will emerge for ICT equipment.
4. Cloud computing will become more prevalent than data centers.

(ZK: I refer to each question as before.)

As for 2-1, a fair and thorough assessment method of data centers will be developed. A data center should be evaluated from environmental, economic, and societal benefit points of view. Currently, the big emphasis is on economics.

(ZK: Often data centers receive beatings from environmentalists because of large space, power, and water consumption. But there should be a fair and comprehensive assessment because a large data center may be more energy efficient than small data centers and provide essential societal benefits.)

(ZK: Regarding 2-2, an embedded solution is generally more efficient than an open solution. The ICT segment has made significant progress by providing open solutions. But when it becomes really necessary to focus on energy efficiency, it may make sense to develop a system that combines hardware, software, server, storage, and networking together, such as Cisco’s unified computing. Within the system, some proprietary protocol may be used, but if IP is used to interact with the system from outside, there should not be a big problem.)

As to 2-3, servers and other ICT equipment will become more self-managing and self-healing. That will cut down human intervention in case of minor to medium-size problems.

(ZK: This may be implemented with redundancy and by moving virtual machines from a faulty server to another to continue processing. Cloud computing does some of this already, but more progress will be expected.)

Regarding 2-4, I already discussed this above. But this opinion was expressed by Ron Hughes of the State of California. His actual comment was that state and local governments will no longer build new data centers but use clouds. (ZK: The US government has indicated this as well, and it also applies to the private sector.)

(ZK: As an IT guy, I welcome the shift in focus from facilities to IT. However, we need to move farther to incorporate data, such as server health and application status, from the higher layers of ICT equipment. They are currently monitored by system management tools. Information like where each application runs and what status each application is in would have an impact on data center operations. More needs to be done in this area.

When I first heard the term private cloud, I did not get it. You do not outsource computing or storage resources but use your own. Even if you do not use some server or storage, you still keep that server and storage in your building, and that costs you money. Something is missing here. Fundamentally, public and private are different beasts, although they may use the same technologies. In a way, with the emergence of private cloud and in contrast with public cloud, the notion of cloud computing has become more crystal clear.

Let’s look at this from a different point of view. When you look back at the progression of your data center in terms of energy efficiency, you used to have one application for one dedicated server to make sure that application runs reliably and securely. Then came virtualization. Virtualizing your applications enabled you to host multiple applications on a single server, consolidating multiple servers to one for higher server utilization. Before virtualization, the server utilization percentage tended to be in the low teens, and multiple servers were required, raising power usage and costs. The next phase is automation of control. A bunch of virtualized servers do not form a cloud, but automation does. With proper management and automation, your virtualized data center becomes a cloud data center.

While I was thinking about this, Greg Ness, Chief Marketing Officer, Vantage Data Centers, sent me a link of a panel he moderated on this very topic. His session is listed at the bottom of the page. The video runs a little less than 30 minutes. Usually, you cannot do much in 30 minutes, but Greg covered several topics, including the definition of a cloud, public vs. private clouds, security, economic issues, and the future direction of IT as it relates to clouds. When you pack so much into a short session, the audience usually loses sight of what’s going and does not get much out of it. But Greg did a good job and moderated the panel to express very useful information.

The following were on that panel:

Moderator:

• Greg Ness, VP Marketing, Vantage Data Centers

Panelists:

• James Barrese, CTO, PayPal
• Winston Damarillo, Cofounder and CEO, Morphlabs
• David Nelson, Chief Strategist, Cloud Computing, Boeing
• Paul Strong, CTO, Global Field and Customer Initiatives, VMware
• Don Pickering, CEO, OneOcean

I will summarize the panel discussion with my comments below. If you want, you can watch the video here. Greg’s session is listed at the bottom. Unless indicated specifically by “ZK,” the following opinions are the panel’s.

Greg started the session by quoting Gartner’s observation that cloud computing is now moving from the height of inflated expectations into the trough of disillusionment in its hype cycle. (ZK: Gartner’s analysis was published in late 2010. The hype on cloud computing started with public cloud and software as a service (SaaS). In a way, public cloud is IaaS [computing first followed by storage]. Then came private clouds and platform as a service (PaaS). In 2010, neither private clouds nor PaaS were given much attention, though.)

The definition of clouds: The first topic was the definition of cloud computing. (ZK: In many panel sessions on cloud computing in the past, the discussion ended here.) Cloud computing features include fast provisioning, pay-as-you-go, and instant scale up/down. The consensus of the panel is that cloud computing is not a new technology but a new business model for the delivery of IT services. (ZK: No disagreements and so far so good.)

Public vs. private clouds: The next question was the difference between public and private clouds. Simply put, private clouds are an attempt to mimic public cloud technologies and the operating model, which are placed outside the enterprise, inside the enterprise. What private clouds could change includes faster provisioning and instant capacity adjustment. For example, the provisioning time was cut from, say, nine months, down to an instant. Once a new operating model is adopted and it works better than the previous one, no one wants to go back to the earlier model. The economic side plays a role in the adoption of private clouds as well. Add to this the process changes due to the new operating model, which makes both developers and guys in charge of architecture work better, and you understand why enterprises are embracing private clouds.

(ZK: In this segment, they did a good job of listing the merits of private clouds, and most people probably agree with their opinions.) Now the discussion got more interesting and covered why and how enterprises move their computing from public clouds to private clouds, as Zynga recently did. (ZK: This is something new to me, and I found their opinion very interesting.) The bottom line is ROI and the utilization of the assets in question. If your utilization of the asset is high, you should own it. If not, you want to outsource it. A good analogy is the ownership of a car. Unlike city dwellers, most suburbians like me own their own car, mostly because they use it daily. However, city dwellers can use other means of transportation and may not use a car daily. So it makes sense for suburbians to own a car and for city dwellers to take a taxi when needed. This is very reasonable and easy to understand.

Unfortunately, our life is not that simple. How do we handle spiky load increases, as in the case of disaster recovery or training, which may not happen very often? That is why we need a hybrid approach that uses both public and private clouds. The panelists advised using private clouds for predictable loads because it is less expensive to do so and using public clouds for spiky and dynamic loads. (ZK: This sounds good but here’s a problem. When people talk about a hybrid cloud, they talk as if virtual machines (VM) running on a private cloud can be easily and seamlessly moved to a public cloud. But as far as I know, it is not that simple. VMware solutions are dominant in the enterprise market, and Amazon’s AWS solutions control the public cloud segment. There are a few variants of cloud file formats, and those used by VMware and Amazon AWS are not compatible. Unless the file formats are translated, moving VMs between public and private is mere theoretical talk.) That is why one of the panelists mentioned the need for cloud standardization.

Another point that was interesting was the simplification and specialization of computing patterns. In a typical large enterprise, there are something like 6,000 applications. Each application needs to be:

• sourced
• designed
• integrated
• deployed
• managed
• monitored
• archived

To do this massive task, for the past 25 to 30 years, 80–85% of the cost has been dedicated to software and those who manage software. Cloud changes this. For example, companies like eBay and PayPal have only a few patterns, which cuts the cost of maintaining them. When you do this, your opex primarily becomes energy, and the issue becomes where you run your loads and how you optimize the power/infrastructure ratio.

At this point, Greg interjected an interesting fact drawn from his discussions with area experts. When your power consumption is less than 500 kW, it makes sense to outsource your computing to a public cloud. But if your consumption goes over 500 kW, it becomes more economical to have your own infrastructures (i.e., private clouds) so that you can tune them for your computing needs.

Security: (ZK: In almost any cloud discussion, security is mentioned as the #1 inhibitor of the adoption of clouds.) Is cloud security a solved matter? One of the reasons some people prefer a private cloud is because it appears to be more secure than a public cloud. Cloud security really comes down to data security. Cloud security needs to address regulatory requirements according to geographic locations. But they are being addressed. Cloud needs to be more secure. It is like spreading peanut butter over a slice of bread. If you spread it, it gets thinner. If you concentrate your security effort on one spot (a cloud) by an expert that does this as their core business every day, rather than spreading it over a mass (many organizations whose business is not security), security works better on the concentrated one spot.

The future of cloud: Some vertical clouds with specializations like financial, health care, specific compliance requirements (e.g., aviation), and government data will emerge. Who will dominate the cloud market, big companies like Oracle or small startups? So many different patterns are required to satisfy the variety of users. And one company probably cannot accommodate such a diverse set of requirements. For that, open-source clouds, such as Openstack, need to be taken into consideration. Although Openstack may be only 80% completed, it supports multiple technologies, such as hypervisor, work processes, and patterns.

In the future, when an enterprise user indicates what their core business is, a service provider like VMware will provide everything that is not strategic to its business and let the company do their business at their core. PaaS will be a differentiator when both SaaS and IaaS become commodities with little differentiation. An ecosystem will be formed around PaaS with specific IPs.

In 5 to 10 years, a lot could happen, such as development of massive amounts of memory/storage, low-power servers (e.g., 10 W computing), software-defined networks, support for multiple devices and platforms, global software development collaboration via pipeline, and standardization. These will change the way software with PaaS is designed and deployed, for sure.

(ZK: Even if you have read this far, I still recommend watching the video because I omitted some points and paraphrased some nuances. The video has a lot of subjects that are current, and you can understand where cloud computing stands at this point. Moreover, I sometimes moderate panel sessions, and I think I learned a lot about how to do that from this panel.)

Some regard social networking systems (SNS) as timewasters, but many others, like me, take them seriously. I am not sure if the current shape and form of SNS will continue to exist for many more years, but certainly mass participation and sharing of a vast amount of information in an elastic fashion will continue in one shape or another for many years to come. This means we need infrastructures to handle such requirements. I will try not to be too analytical, but these are two of those requirements:

1. Instant (by the second) increase/decrease of traffic of massive proportions with dynamic fluctuation
2. A vast amount of structured and unstructured data processing

I am sure there are more, but let’s just make things simple by considering only those two. The first states the necessity of having a powerful infrastructure that can process an unseen amount of information. Because of dynamic load shifting, a static infrastructure is not a good solution. In other words, we need an infrastructure to satisfy somewhat conflicting requirements. One solution is to oversubscribe each and every component in the infrastructure. However, most times you cannot accurately predict the level of demand, especially for SNS applications. Because you cannot prepare an infinite amount of resources, this solution would be very hard to implement. Also, as stated in the second point, a vast amount of multiple formatted and unformatted data must be processed. On top of that, those data are coming in mass in real time. How do you handle that?

In a recent cloud computing meet-up held at Microsoft’s Silicon Valley office, there were two talks on accommodating these requirements.

The first talk was by Jim Zimmerman, CTO of Thuzi, which is based in Tampa, Florida. According to their website,

Thuzi is focused on making social media count.

Jim Zimmerman

To that end, they provide infrastructures and tools to assist clients’ SNS-based marketing campaigns, which might be hosted on Facebook or other sites. Their tools provide a solution to the above two requirements.

Jim said that the techie population in Tampa is much smaller than that of Silicon Valley, and when they hold a meet-up, they can do it at a local Panera Bread Restaurant. In contrast, a good-size auditorium at the Microsoft Research facility was full of techies. I must confess that I could follow most of his talk, but here and there I was lost because I have not touched code for some time. But the problems he articulated were clear. They have a solution for both business and consumer applications.

A set of problems in dealing with business applications.

The problem statement is an expanded version of what I described. For this, their solution is here.

Command query responsibility segregation (CQRS) is not a very familiar term. Jim elaborated it further in the following architecture slide. Basically, it is a complete separation of UI and backend. The same philosophy was discussed in the subsequent talk by Brian Aker.

The consumer application has its own problems, as follows

And their solution for that is as follows.

They also have an application that manipulates photos. Their earlier version sent everything to the server for processing, but it took too much time. In the current version, they process the photos on the client side and send the processed photos to the server. This is summarized in the following slide.

The computing and infrastructure requirements need very different thinking to accommodate SNS applications. Public cloud is an ideal platform to accommodate unknown bandwidth because you can increase your computing and other resources on-the-fly. If traffic subsides, you can decrease the resources accordingly. So for you, it is good to save energy, but how about the provider? In theory, a public cloud provider has many customers running various loads simultaneously. Some may increase and others may decrease traffic at a given moment. So in the end, everything balances out and energy is not wasted. Intuitively, that seems to be true. But it really depends upon many parameters, including each cloud site’s configuration and customer mixes. I think the jury is still out on this.

I cannot generalize their architecture but it seems that for scalability support, UI and backend processing should be separated to increase processing power, regardless of how light or heavy traffic may be. In the cloud environment, UI and backends can be anywhere. They may not be stationary at one physical location, and that makes the whole design very complex. Brain Aker addressed that problem in his talk.

The last slide contains the relevant reference information.

The conference was divided into very techie, hands-on workshops on actually hacking something on-the-fly and not so technical ones. It has been so many years since I hacked code, and I did not catch up with the conversations exchanged in the workshops, so I mostly attended keynote speeches and not so technical sessions. Other cloud conferences offer many different subjects, including markets and businesses, to attract a large audience. Yes, those things were discussed during the not so technical sessions, but there was something that was different from other nontechnical sessions in other conferences. I cannot say it very well, but my analysis is that the not so technical sessions were more technical than the general discussions of cloud computing in other conferences. Even though I thought I already knew the basics of cloud computing, many of the talks were geared towards technical folks, and I enjoyed them very much. In this blog and others to follow, I plan to summarize these discussions and inject my thoughts. I actually have written a blog on open source cloud computing.

In this blog, I would like to concentrate on Gordon Haff’s talk on cloud computing.

Gordon Haff of Red Hat

The stuff he covered was pretty basic, but I guarantee that you will appreciate what he said. It will give you an even better understanding of cloud computing. Here goes.

Gordon made five points about cloud computing.

Point 1

What is the fundamental difference between cloud computing and traditional enterprise IT? The next slide says it all, and I guess no explanation is necessary. Many people have made similar points, but this presentation summarizes the key points well. Remember that this is about public cloud. Public cloud went far in cultivating the cloud market, and private cloud is an attempt to duplicate that success in a private environment for enterprises. In a separate panel discussion, Nati Shalon of GigaSpaces summarized it very well. He said public cloud was all about productivity, while private cloud was about control. As the slide shows, public cloud enhances productivity in terms of agility and resource conservation (both cost and time). What is missing in this slide is control and security. Security is addressed in a following slide.

The control issue is sketched out in the following slide. Public cloud is great when we can ignore the issues listed on the left side, such as risk, compliance, guarantee, and differentiation points. In other words, enterprise adoption requires those issues to be resolved or at least for a compromise between risks and productivity to be reached after careful consideration.

Point 2

Cloud is fundamentally a computing model for data created by mobility, which is increasingly adding new types of data, such as context-aware data like locations and environments. Of course, stationary sensors placed at strategic locations of buildings, data centers, and smart grid systems also add to those data. The data that are generated, collected, and accumulated form Big Data, which is becoming a place for treasure hunting. Cloud is becoming the computing of choice for Big Data because of its traits, listed next to the cloud computing arrow in the following slide.

Of course, open source plays a large role in this, but even with a bunch of open source solutions for cloud computing, we still have very limited standardization. See my old post.

Automation, scalability, and self-service are all necessary for coping with this huge amount of data, or even to make heads or tails of it.

Point 3

Openness is a great concept for ICT practitioners. In a given enterprise, there are usually several computing entities invested in separately by different departments. To date, most of them are operated in a silo, independent of each other, and valuable data and information are not shared across departmental barriers. If the barriers are removed, all the investments, whether internal or external, can be used for the benefit of the entire enterprise.

This can be accomplished by open cloud. In the following slide, Gordon showed what is required for open cloud. Although I agree that open source is a big factor, it alone will not accomplish this. Take Linux as an example. In addition to a few dominant distributions, there are quite a large number of other Linux distributions, as indicated here.

Each added its own utilities and libraries to the relatively small Linux core. If this is any indication, open source cloud will diverge in many directions with the small, standard core.

But don’t think I am against open source. On the contrary, I am all for it. Points made in the two following slides are self-explanatory.

The existence of the community and ecosystems is important. The entire area of cloud computing is very wide and deep. No single product or service vendor can address all the issues and needs of the market. Therefore, it is very important to cultivate and grow the community and ecosystem as in other market areas.

Point 4

Security remains the largest inhibitor for enterprises to adopt public cloud. In the next slide, Gordon compared enterprise computing and cloud computing. Humans can perceive danger if it is visible, but we tend to ignore invisible threats because we cannot see them. We usually worry about perimeter security because an intruder who cannot pass the iron and cement gate cannot do any harm to our computing infrastructure. That is true, but in the age of ubiquitous connection, simply guarding the perimeter is not enough. We need to pay attention to network and other computing resource security.

Many analysts and experts say that the security available on cloud and hosting providers’ premises is usually much tighter than that of the typical enterprise. The reason given is that those who provide services focus on secure delivery because that is their core value proposition. On the other hand, enterprises worry about security, but securing their computing infrastructure is not their core value of business.

Gordon said that securing everything at an enterprise might give it some security and risk mitigation, but the resulting slow response to ever-changing circumstances might cause it to lose business opportunities.

Point 5

Finally, the following slide is great. Many people have discussed the progression of computing at enterprises. The following is a good way to understand it. Many people are still scratching their heads over the difference between a virtualized data center and a cloud data center.

Many people appreciate that virtualization revolutionized computing at data centers. It allows us to run multiple computing units (virtual machines, or VM) on a single server, cutting the number of necessary servers and increasing server utilization. At this stage, each server is independent and managed by a separate policy. This is a virtualized data center, but it is not a cloud data center. The next level up is a data center with holistic control and management of those virtualized servers with the same automation and policy.

Gordon further refined this model by adding applications and data management. This is where we are headed for. The current issues and interests are in applications (platform-as-a-service, or PaaS) and data.

Cloud people are busy with enhancing cloud functionality and other features like agility. I wonder whether cloud efficiency in terms of less resource consumption will become an issue anytime soon.

It was simple in the enterprise world. There was a main business organization with a handful of supporting departments like HR, accounting, legal, and facilities management. There was no doubt that the business organization was king. It ruled the enterprise without question. Then computers came to the enterprise. At the beginning, computers were subordinate, simply supporting business. As computer technologies advanced, it became necessary to form an independent organization called IT. Even so, IT was subordinate to business. Business dictated and IT followed.

It is hard to say when, but at some point IT got so powerful that business could not tell it what to do. Or more accurately, business was still boss, but IT became less responsive to business needs, and not just to requests; IT did not and could not respond to the changes taking place around it. IT was supposed to support and accelerate business goals, but it became a barrier to business. It became so bad that people said IT was the place where big, important projects went to die, according to Tim. For some time, business was frustrated with IT but did not have the means to bypass it. Then came the era of cloud computing and commercialization of IT. Business secretly formed a shadow IT department, like a shadow cabinet in the UK, and started to bypass IT whenever possible. Who can blame non-IT folks who need IT services yesterday? If it is going to be months before IT can satisfy my needs when I need it now, I will bring in my own gear or outsource the services. The big difference now is that we can do it if we want. It was not possible to do so only a few years ago. Tim said some CIOs now realize this and are working to face this straight, but many CIOs still think the old and traditional way of running IT departments is appropriate.

There is no real department for shadow IT. In a way, any business or non-IT staff who needs IT services can virtually form a shadow IT group and outsource their needs to cloud and mobile computing. If there were really one physical shadow IT department, it would be easier for the real or traditional IT department to confront it and take back control. But this shadow IT group is like guerrilla warfare. There is no particular place the group shows up. It appears where there is a real need for IT services, gets them quickly, and then may disappear. As far as I know, Tim is the first person in IT to admit that the blame is on IT for bringing this situation on itself.

This may not be a good analogy, but at the same conference Pascal Finette of Mozilla gave a keynote speech on open innovation. His theme was that opening the barrier could accomplish even greater results. Maybe it is a stretch, but I am saying that the IT department should be more OPEN to what their internal customers want and work with them for the entire company. Tim said that CIOs in this new era should take a very hard look at their core capabilities, review their portfolio of services, and decide what should be retained and what should be outsourced. This is a hard thing to do because it may mean downsizing the IT department. The CIO and the IT group need to have heart-to-heart discussions among themselves and with their internal customers.

It is easy to propose this at a high level, but how do you actually implement it? When it comes to cloud computing, the number of popular services like Salesforce.com and AWS is low, and it may be easy to evaluate the usefulness of each. But if IT approves a “bring your own device” (BYOD) policy, the sheer number of variations could be a problem. I would like to ask Tim about this. Maybe he has already addressed this in his blog.

Tim advocates that IT take the initiative to evaluate current services and gear before its internal customers come asking for support and approval. It would be great if, when a customer asks for services or new technologies, IT is ready to embrace the request and seamlessly integrate it into the current portfolio of services. That is a new value to IT and will definitely increase their importance in the company. There is going to be new pressure on IT. IT needs to study the market trends and new services and technologies daily and constantly. This alone would produce a ton of work. So this makes it even more important to review what they currently have and remove services of low priority from their portfolios. If some services can be easily outsourced without losing important elements, such as security, they should be.

For example, if IT studied Salesforce.com well in advance of a customer request to incorporate its services for business, it could accommodate the request with strong support. Moreover, because IT is probably the only department that interacts with most, if not all, departments in the company, it can facilitate communication between departments. For example, if IT understands how marketing and sales operate, it could avoid sending conflicting messages to the same customer from marketing and sales, which is very embarrassing, according to Tim.

What if IT provides virtualization? Would it be enough to prevent internal customers from resorting to outside clouds? Virtualization and cloud are two different things. I caught Tim after the interview to get more details on this, but that is a story for later. For now, let’s say virtualization is not cloud, and cloud is more than that. Tim said that the three pillars of cloud computing are economic value, flexibility, and responsiveness. In most discussions, economy is considered the key value for cloud computing, but Tim said it might not be the most important factor. Sometimes cloud computing may cost you more, but you may want to adopt it for its flexibility and responsiveness. To sell cloud computing to your CFO and CEO, you should be able to make each point by giving a concrete example and the savings associated with it. For example, scaling from 50 to 1,500 servers over days and weeks could only be accomplished via cloud computing; no traditional methods accommodate change on such a scale. Moreover, you should be able to convert this to $ to explain it to the CFO and CEO.

Is virtualization a necessary process before adopting a cloud service? Tim said that the adoption of virtualization is probably in the 30% to 40% range, less than the 50%-plus estimated by IDC. If you have not virtualized your data center yet, you need to take a holistic view rather than simply considering virtualization. Looking solely at technology (virtualization) is not right. After all, virtualization maximizes server utilization, but other things—like maximization of the organization’s resources and processes—are also very important. So, according to Tim, it is OK to bypass virtualization and adopt cloud services by paying attention to the whole picture.

What about organizations that have already invested in hardware? It is easy to talk about startups, which have very little IT gear, but larger companies, even small to midsize companies, may have a hard time moving to a new paradigm. Tim’s answer was clear: whether it’s hardware or software, everything will be replaced at some point. The lifespan of server hardware is about four to five years, and software applications must be upgraded or replaced at some point. That is the time to take a hard look at what to do from several points of view: economy, technology, process change, and the whole organization.

Finally, security is the number one inhibitor mentioned in many research results. The perception is that only larger companies need to consider both security and regulatory conformance and that small startups can adopt cloud computing readily. Tim said both considerations are required in any business of any size. So that kind of argument does not have merit. Also, he said, in many instances cloud providers’ data centers and security measures for other elements are better than those of companies of many sizes and kinds. In any event, if you worry about data security, you need to weigh convenience and ease of use with security risks. Having data at your own data center does not make it more secure than the data hosted somewhere in clouds.

There is a lot of talk about cloud computing, but sometimes people duck the hard questions. I will catch Tim in the future to discuss some of mine:

• How do you transition from virtualization to cloud computing?
• What are public, private, hybrid, and federated clouds?
• Is there such a thing as private cloud?
• Hybrid clouds? What about interoperability of VMs between private and public clouds?
• What new requirements, such as ID management, do federated clouds bring?
• What are green clouds? Can you quantify the greenness of clouds?

So stay tuned!