The Silicone Whisperer

As we come to the close of 2009 and the larger close of a decade, I remind all of you out there to test your backups. With recent events like the loss of flight data at Shaw Air Force Base and subsequent realization that backups were not working, it is more important than ever to test the integrity of your backup, archive and restoration platforms. It is no secret that enterprises have slashed IT budgets over the last few years, and some of this pruning came at the expense of backup platforms and personnel. In many of these enterprises, no one has stepped up to verify that the critical data backups (if they are even backing this data up) are valid. I implore you to not be one of the negative statistics. Let’s take a look at the holy trinity of Disaster Recovery and Business Continuity.

Backup

Once a company has identified the business value of systems and data, they typically assign a risk value to losing those. This typically sets the wheels in motion to get a backup system in place. Backup, the very first step in Disaster Recovery and Business Continuity planning, is the base upon which you will build your strategy. Without the backup, there is nothing to archive nor restore later. Backing up typically entails duplicating data onto a secondary medium which acts as a safeguard against primary storage failure. This can be something as simple as a disk to disk (D2D) replication to a second storage system, or as complex as an NDMP stream across a fabric infrastructure to tape libraries waiting to write the data to magnetic tape media.

Archive

If we then take those tapes, on which we backed up our data, and move them to a secure remote storage facility, we have now archived that data. The main difference between a backup and an archive (although some hybrid solutions out there blur the lines a little) is that backups can be on-line or live while archives are typically off-line and stored somewhere other than the physical facility that the primary data storage resides. A disk to disk solution can be an intermediary to a full fledged tape backup and archive solution with the ability to restore data locally in a very rapid fashion. With an archive, you typically have to request the appropriate data set (on tape media) from the off-site facility, load it into the local tape library and restore the data from the media. This is essential for comprehensive Disaster Recovery and Business Continuity plans. If your primary physical location becomes unusable due to natural or man made disaster, you can relocate to a new facility and restore all data from tape media. The down time may be longer versus a local backup solution, but the down time is not permanent. Failure to have an adequate off-site data archival solution can result in loss of all data and permanent down time.

Restore

The end result of data loss or catastrophic storage system failure is a restore from the appropriate media. This final step should always be prefaced with several tests to validate the integrity of the data residing on the local or archived backup. When there is a loss of data, a local backup can be tapped to restore the lost data, or a data restoration from archival media can be made. Either way, the lost data is replaced with the last known good copy, and business continues as usual. Every successful restoration hinges on the validity of the data that was backed up in the initial stage. If any part of the backup and archival process failed, there is a good chance that the media will not be valid for restore. This issue can be compounded when your backup, archival and restore system does not properly alert an administrator to an error in the process. This is why tests of the backup media are so critical.

There are a plethora of backup, archival and restoration solutions on the market, and they vary greatly in cost and complexity. Some include features that many enterprises will not need, and some are missing in areas where a more robust solution is often warranted. It is most imperative that you first plan your Disaster Recovery and Business Continuity scenarios, then select the best platform to fit those. While one enterprise can get away with simple storage to tape archival, more complex enterprises will want array level replication to a hot site with local D2D cached copies and a full fledged tape archival solution. Adequate cost analysis and risk assessment should point you in the right overall direction. Avoid leaning on a particular vendor and seek the professional assessment and recommendation of a DR or BC adviser.

The Ubuntu team has decided to take a different approach to development for the next Long Term Support (LTS) release of Ubuntu. Lucid Lynx, which is slated for an April 2010 release, will benefit from these changes through increased stability, more focused agile development and a tighter synchronization with the main Debian development cycle. According to the Ubuntu LTS Wiki, the overall aim of an LTS release is to be more enterprise focused with increased hardware compatibility and more overall testing. What is not intended for an LTS release is cutting edge package inclusion nor a hot of new features. Because the core focus of an LTS release is stability, the user base should appreciate the level of refinement that will go into the release and not fret over the lack of ‘cool new features’ that often accompany non-LTS releases.

So how does Ubuntu plan to ‘reign in the sprawl’ with respect to development of Lucid Lynx? I believe Matt Zimmerman, technical leader of the Ubuntu project, said it best on his blog.

Merge from Debian testing

By merging from Debian testing, rather than the usual unstable, we aim to avoid regressions early in the release cycle which tend to block development work. So far, Lucid has been surprisingly usable in its first weeks, compared to previous Ubuntu releases.

Add fewer features

By starting fewer development projects, and opting for more testing projects over feature projects, we will free more time and energy for stabilization. This approach will help us to discover regressions earlier, and to fix them earlier as well. This doesn’t mean that Ubuntu 10.04 won’t have bugs (with hundreds of millions of lines of source code, there is no such thing as a bug-free system), but we believe it will help us to produce a system which is suitable for longer-term use by more risk-averse users.

Avoid major infrastructure changes

We will bring in less bleeding-edge code from upstream than usual, preferring to stay with more mature components. Where a major transition is brewing upstream, we will probably opt to defer it to the next Ubuntu cycle. While this might delay some new functionality slightly, we believe the additional stability is well worth it for an LTS release.

Extend beta testing

With less breakage early in the cycle, we plan to enter beta early. Traditionally, the beta period is when we receive the most user feedback, so we want to make the most of it. We’ll deliver a usable, beta-quality system substantially earlier than in 9.10, and our more adventurous users will be able to upgrade at that point with a reasonable expectation of stability.

Freeze with Debian

With Debian “squeeze” expected to freeze in March, Ubuntu and Debian will be stabilizing on similar timelines. This means that Debian and Ubuntu developers will be attacking the same bugs at the same time, creating more opportunities to join forces.

He goes on to talk more about the set of tools that the developers will be using to track their progress and the use of agile development methods to make that easier. The developers will also be posting their progress publicly so that the community can follow along. In essence, the final goal of this LTS release is to gain the confidence of IT departments who will be deploying Lucid Lynx and maintaining it for a period of years thereafter.

As Ubuntu seeks to become the “open platform of choice for almost all use types whether on netbook, notebook, desktop, server, embedded device or wherever people compute,” Mark Shuttleworth has decided to step down from the perch of Canonical CEO. Replacing him will be current Canonical COO Jane Silber. Mark has decided to focus more on his “passion of product design and development.” Jane Silber, a former VP at General Dynamics, has much experience in dealing with very large customers. This means that Canonical will be leveraging Jane’s experience in this market to woo the largest of customers.

So what does this mean for Ubuntu and Canonical? Well, as Mark Shuttleworth put it, this shift “will bring about a clearer separation of the role of CEO of Canonical and the leader of the Ubuntu community. It will be two different people now, which [he thinks] will be helpful in both achieving their joint and individual goals more quickly.” I find this move to be highly strategic and very well executed to promote the growth of both the open source community backed distribution and it’s corporate support arm.

When you are dealing with the open source crowd, passions are high in all sorts of areas. One thing that would ring foul to most in the community would be to have a ’suit’ running the community project. I am in no way saying that Jane is a ’suit’ but being a former VP of a large corporation would definitely paint that picture. Mark, despite being the CEO of Canonical up to this point, was more than likely viewed as a geek ‘who happened to run a business.’ To push the role of CEO of Canonical along with Leader of the Ubuntu Community onto Jane would have been disastrous from the perspective of the community. By retaining the leadership position with the Ubuntu Community while passing the corporate responsibilities for Canonical onto Jane, Mark was able to win the hearts and minds of people on both sides.

So how does this affect the Ubuntu Community? For most things, there will be little to no impact as Mark is still the leader of the community and the project will move forward as usual. The distribution will gain from Mark pouring all of his passion into improving the user experience at all levels. All in all, Mark will simply be “spending more time on the areas that interest [him] the most and where [he feels he] can do the most good.” I expect some really great things in the UI department over the next few releases. This will please the desktop and portable crowd more than anything.

On the corporate side, I believe a renewed focus on large enterprise customers will yield the revenue that Canonical has yet to find. Profitability has remained just out of reach, but the company has persevered regardless. It will take a keen and focused CEO to supplant Red Hat as the de facto Linux distribution in the enterprise, but if I had to pick the most likely contender to do so, it would be Canonical/Ubuntu. The keys to dominance in the enterprise operating system arena are excellent vendor support, a solid well defined road map, a deep ecosystem of system architects, designers and developers, stability and flexibility within the infrastructure. I believe that Jane will be able to focus on these core areas and develop relationships with many Fortune 500 enterprises.

The pervasiveness of the Ubuntu Linux distribution is such that it is considered ubiquitous at this point. Of those who are not familiar with the Linux universe, most have at least heard of Ubuntu and to them it means Linux. This is a great perception to have when developing image and branding. Ubuntu already has deep credibility within the open source community and many SMBs. The last step for Ubuntu is to carry that credibility into the Fortune 500 market. It will be a difficult uphill road to dominance in the enterprise data center, but much like the ant eating the elephant, it will be accomplished one bite at a time.

As the fog slowly lifts with respect to defining cloud computing, enterprises must now decide where the value proposition lies within the platform. For most large enterprises, the IT department typically offers services internally to various departments. These services are mostly at the infrastructure level such as maintaining existing servers or provisioning new servers. Some enterprises allow developers to build out their own platform while more organized enterprises keep tight control over the platform, only allowing developers to work within a tightly pre-defined environment. While this is definitely the preferable way to reign in developers, it still isn’t the most optimal way to develop in house applications.

Deploying a platform as a service to developers, who in turn develop software to be deployed as a service to internal departments, is the best way for a large enterprise to leverage cloud computing. This takes two levels within the cloud computing pyramid and uses one to bolster and deliver the second. The IT department will ultimately be responsible for the three layers beneath platform as a service, but more robust management tools are making that easier as time goes along. Ultimately, automation and orchestration will become the panacea for IT services delivery within the enterprise.

Automation will allow for the infrastructure to scale up and down dynamically without system administrators having to manage the process. This is true elasticity within the enterprise. The next level, orchestration and life cycle management, is the magic dust that makes this whole technology worth its weight in gold.

Imagine taking the workload manager within an application and making it aware of the cloud it resides in. It now knows that there are more resources available within the cloud, and that these resources can be tapped on demand via additional virtual instances. As a major workload hits the application, the work queue starts to back up. The workload manager sees this and begins to spin off another virtual instance down at the virtual infrastructure level. This instance comes on line, and half of the workload is now routed to that instance. Now another surge of work hits the work queue and again the queue begins to back up. The process repeats, and another virtual instance is spun up and comes online. Now the workload manager can route work to three instances.

This process happens several more times, and now we have ten virtual instances with the workload manager load balancing across all instances. As time goes on, the workload begins to reduce and one instance is no longer needed. The workload manager signals the virtual infrastructure level of this and one instance is destroyed. THe workload decreases some more, and now another instance is no longer needed. Once again, it is destroyed at the indication of the workload manager. This process continues several more times until just one instance remains.

The elastic nature of the cloud allows for the rapid creation and destruction of virtual instances. The software which has been developed to run on this platform is intelligent enough to analyze workload and direct the underlying virtual infrastructure. The ability to rapidly provision and destroy identical virtual instances is automation. The ability to make this automation happen dynamically based on workload is orchestration. So where does the true value lie in cloud computing?

That is a heavily loaded question because there is value in almost every aspect of cloud computing, depending on the size and complexity of your enterprise. If I had to pick one thing, however, and say that it provides the true value for large enterprises, I would say that it is orchestration. Orchestration truly makes applications dynamically elastic.

So what should you take away from this article?

I would say that as the various virtualization and cloud vendors jockey to out do one another with frivolous features, you and your IT staff take a close look at automation, orchestration and life cycle management capabilities in their respective products. Ensure that your developers can hook into a solid API that allows for true orchestration. The process should also be fairly invisible to the developers. They should not have to understand deeply technical concepts for them to spawn and kill virtual instances as the application needs them.

Ideally the platform should also be fairly robust when it comes to provisioning hardware. Once you have established a solid architecture with templates for the virtual instances and baselines for all hardware in the cloud, addition of physical nodes should not me much more complicated than racking and cabling the hardware. The hardware should then be able to PXE boot itself, authenticate, and start auto provisioning itself into the cloud. The idea is less hands on management and more intelligent automation.

Some great news came out of Sun Microsystems yesterday with the release of VirtualBox 3.1.o. This is Sun’s virtualization platform, which has been at the core of many of Sun’s newest technologies. What is great about VirtualBox, aside from being a professional quality hypervisor based virtualization solution, is that it is open source.

According to Sun, the newest major featured added are:

• Teleportation (aka live migration); migrate a live VM session from one host to another (see the manual for more information)
• VM states can now be restored from arbitrary snapshots instead of only the last one, and new snapshots can be taken from other snapshots as well (“branched snapshots”; see the manual for more information)
• 2D video acceleration for Windows guests; use the host video hardware for overlay stretching and color conversion (see the manual for more information)
• More flexible storage attachments: CD/DVD drives can be attached to an arbitrary IDE controller, and there can be more than one such drive (the manual for more information)
• The network attachment type can be changed while a VM is running
• Complete rewrite of experimental USB support for OpenSolaris hosts making use of the latest USB enhancements in Solaris Nevada 124 and higher
• Significant performance improvements for PAE and AMD64 guests (VT-x and AMD-V only; normal (non-nested) paging)
• Experimental support for EFI (Extensible Firmware Interface; see the manual for more information)
• Support for paravirtualized network adapters (virtio-net; see the manual for more information)

The first of these new features, ‘Teleportation’, is HUGE. The ability to migrate a live virtual machine really makes downtime a think of the past and the true multi-platform nature of VirtualBox (Solaris, Linux, Mac and Windows), allows for this migration to take place across operating systems, hardware vendors, processor architectures and even from servers to clients.

Allowing restores from arbitrary snapshots makes for a pseudo revision control system for virtual machines. This is a definitely a feature that will come in handy for developers and testers.

2D video acceleration will definitely help on the virtual desktop front. Flexible storage attachments allow users to more accurately mimic physical hardware for certain systems, and changing a network attachment type on-the-fly can help with virtual network testing. The rewrite of USB drivers for OpenSolaris brings the platform closer to plug and play functionality.

A performance increase for PAE and AMD64 guests is always a welcome thing as is the foray into EFI support. Naturally, the paravirtualized network driver will drastically increase networking performance for guests.

I would say that Sun has made some major advancements overall in their flagship virtualization platform. Having used VirtualBox on several machines with Unix, Linux, Mac and Windows in my lab, I can definitely attest to the speed and usability of the platform. The added ability to migrate my virtual machines across all of my hosts is a very welcome addition. Paired with SunRay Software 5, I can see the newest release of VirtualBox serving as a solid foundation for a Desktop VDI platform.

As a preface to the series of articles I will be writing on the Value Proposition and Business Cases for Cloud Computing, I wanted to discuss the layers below and within the cloud. It is important to understand what each of the layers is composed of, what the intended function of that layer is, and how these layers interact with each other. By simplifying the cloud computing concept into layers, it is easier to define the roles within the overall structure and explain where your business fits into the model.

Let me start by introducing a graphic I whipped up real fast. In the diagram, you see that the cloud is basically a pyramid of technologies stacked in a certain order. I believe that a pyramid best illustrates the conceptual ’size’ of each layer and it also conveys the message that each layer is based upon the previous layer. So conceptually, you understand that the hardware is the foundation and the widest layer. Software as a Service (SaaS) is the capstone and consequently the lightest layer. Note that this view is from that of an end user who would purchase SaaS from a vendor. For a very large enterprise who develops software internally, the Platform as a Service (Paas) layer would be the capstone for the enterprise. The internal departments that use the internally developed software would see SaaS as their capstone. Also note that the size and position of a layer does not necessarily equate to importance. If you read my previous article on ‘Cloud Servers,’ you would see that the hardware layer is probably the least significant as it is all commodity beyond a certain point.

The Hardware Layer

There isn’t much to be said here that I have not covered in more depth in my previous ‘Cloud Server’ article. I will summarize by saying that the hardware is expected to be inexpensive commodity hardware with no fault tolerance. Redundancy and fault tolerance are handled within the software layer, so hardware is expected to fail. This is completely counter intuitive to most people in the IT field as they are used to procuring fault tolerant servers with multiple power supplies, RAID disk arrays, etc.

The Virtualization Layer

I don’t want to call this the Operating System (OS) virtualization layer despite the fact that this is what is essentially going on at this layer. The physical hardware is being sliced into virtual machines that each have their own small (usually Linux or Unix based) operating system installed. These virtual machines are then collected into pools based on the resources they have. These pools of resources are the key to providing elasticity with respect to server architecture. These virtual machines can be brought online and assigned to a resource pool on-the-fly when the demand on that pool increases. They can then be destroyed when they are no longer needed. The ability to provision and destroy virtual machines on the fly allows a vendor to provide Infrastructure as a Service (IaaS). There are many major players in this commodity layer such as VMware, Citrix, Sun Microsystems and even Microsoft.

The IaaS Layer

The IaaS layer extends the virtualization layer by providing the mechanisms to provision and control the virtual machines in a utility computing manner. By this I mean that a common interface, such as a web portal, or an exposed API allows the end user to build and configure virtual machine templates as needed. The end user can also control when to turn on or destroy virtual machines and define how the virtual machines are networked with each other. If the IaaS is provided in a utility computing model, then the end user is also able to control cost by knowing exactly how much each virtual machine instance costs per minute/hour. Some vendors in this layer also provide storage and database services which are controlled via an exposed API as well. These services are typically billed in a utility computing fashion too. A lot can be done at this layer with respect to automation and orchestration (topics I will discuss in later articles) but the bulk of it is left up to the end user to handle. A good example of a major player in this space is Amazon Web Services with the EC2, S3 and Database services.

The PaaS Layer

The PaaS Layer extends and abstracts the IaaS layer by removing the hassle of managing individual virtual machine instances. This layer seeks to minimize the hassle and complexity in deploying an application in the cloud. A programming platform is presented to the end user, typically a developer at this point, which leverages an API and programming language. Two good examples of this layer are Google’s App Engine and the Force.com platform. With App Engine, Google exposes an API for storage, platform and database and leverages the Python and Java programming languages. Developers are able to write an application and deploy it straight into the cloud in this layer. Back end scalability is handled entirely by Google and the end user does not have to worry about managing infrastructure. The Force.com platform is similar but utilizes a custom programming language called Apex. If you are a large enterprise looking to deploy internally developed applications, this is your capstone.

The SaaS Layer

If you are a Small or Medium Enterprise (SME) or a large enterprise not wishing to develop your own application, the SaaS layer is your capstone. In this layer, even the platform has been abstracted away from you as and end user. You are simply interested in procuring a service, such as email or CRM. The vendor has developed an application and deployed it to the cloud. They handle scaling on the back end, infrastructure, etc. You simply authenticate to the application and use it as needed. Billing can be based on utility or a flat monthly fee. Either way, it is a simple way to get the application functionality you need without incurring the cost of developing that application. More and more enterprises are subscribing to SaaS offerings like Salesforce.com and Sugar CRM.

Putting it all Together

I hope that I have shed some light on the layers that comprise the cloud in cloud computing. Understand that the lines can blur between the layers as vendors within each layer add functionality and features to differentiate themselves from competitors in the space. This model is not concrete, nor does it illustrate every possibility with respect to layering. I am sure as time progresses, these layers will shift, blur or disappear entirely. This is meant more as a guide to understand a concept. In the coming series of articles I will publish on cloud computing, I will delve into greater detail within each layer, show the value proposition, and show how your enterprise can leverage cloud computing to provide a product and service portfolio that aligns with business goals. IT will have to shift from a sunk cost center to a value add business unit that innovates and provides measurable value to the company. Cloud computing makes that transition easier and more structured.

In one of my older articles, I explained how Ubuntu Server has slowly made significant inroads in many enterprise data centers. Ubuntu has been the favorite distribution of many system administrators, architects and various other IT staff. It was these fans of the Linux distribution that brought Ubuntu into their data centers and slowly deployed Ubuntu into ever increasing roles. I have seen Ubuntu Server deployed in roles from simple kiosks and desktops to full blown web server and database farms. Many senior level managers do not even know that key components of their infrastructures are running on Ubuntu Server.

Most data centers use an enterprise backup platform like Symantec’s NetBackup. These platforms are often very complex and extremely expensive to operate. Without a solid comprehensive backup strategy and platform, however, you place your disaster recovery and business continuity in great risk. For those users of Ubuntu Server who do not wish to incur the expense and complexity of NetBackup and other similar platforms, Arkeia Software has released a free version of their popular enterprise backup server for Ubuntu Server. This strategic move lends Ubuntu more Enterprise Credibility while also giving Arkeia more visibility and market share within the enterprise backup market. Lest you think that this is a watered down version of the product line, Arkeia has spelled out what is available to Ubuntu Server users along with the capabilities of the platform.

From the site:

Arkeia Software provides packages specifically designed for Ubuntu distributions, both 32bit and 64-bit architectures, along with specialized agents for the protection of applications and databases running on Ubuntu, such as Oracle, MySQL, PostgreSQL, and LDAP. Ubuntu is also a supported platform for Arkeia Network Backup Disaster Recovery for Linux, which enables bare-metal restores of entire systems.

Arkeia Network Backup: Enterprise Edition for Ubuntu

Arkeia Software provides a fully-licensed, free version of Arkeia Network Backup that is available in the Ubuntu 8.04 LTS software repository. With a few clicks from within the GUI, or just a simple command using the ‘apt-get’ utility, Ubuntu users can quickly deploy a fully-featured backup server. Arkeia Network Backup: Enterprise Edition for Ubuntu offers an advanced network backup solution with the ease-of-use and deployment that Ubuntu users are accustomed to.

What’s Included

• Arkeia Network Backup Server for Ubuntu 8.04 LTS
• 2 Client Agents supporting Windows workstations and desktops, and a vast majority of Linux machines, Mac OS X and BSD computers
• 1 Drive license for disk-based (up to 250GB) or tape backups
• Support via online forums, knowledgebase, and wiki

Learn more about Arkeia Network Backup.

Arkeia and Open Source

Arkeia Software was the first professional network backup solution for Linux and has supported the open source community since 1999. Today, Arkeia continues to provide the deepest and broadest support for Linux and open source with more than 100 platforms supported. Arkeia acknowledges the work of thousands of Linux users who have donated their time and expertise towards the goal of making Linux and open source software a viable alternative.

I received an email letting me know about a new company, Indicee, that will be publicly launching within the next few weeks.

The Indicee story includes a few key elements: a notable team, an innovative product and impressive financing. Indicee is a new arrival on a very competitive landscape but is filling a void left by most BI and reporting software available – allowing users and businesses to avoid the pain of dealing with spreadsheets and mashing data to create reports.

• Indicee provides the quickest and most effective way to create and share reports. Users with no IT expertise can combine data from their business applications and generate reports using Indicee’s innovative cloud-based service.
• The Indicee vision is 20 years in the making. Indicee Co-founder, Mark Cunningham was part of the team that developed Crystal Reports, which grew to become the dominant product in its category. Today, Crystal Reports from SAP’s Business Objects division is the tool of choice for software developers, IT professionals and business intelligence consultants worldwide. Indicee’s vision is to bring reporting and analysis to the “masses” – the Holy Grail that the team was striving for when they first launched Crystal Reports in 1991.
• The track record and reputation of the Indicee founding team paired with the product concept and market opportunity has enabled the company to attract the financing required to fuel growth. Indicee recently completed a $6M initial VC financing, bringing the total raised since inception to more than $8M. Indicee’s angel investors include the original founders of Crystal Decisions. Indicee’s VC investors are Yaletown Venture Partners of Vancouver and Granite Ventures of San Francisco.
• There are three key aspects that make Indicee unique:
• How Indicee accesses data. The near-ubiquity of reporting infrastructure, such as Crystal Reports, provides a virtually untapped and readily-accessible source of data and meta-data that, with Indicee, can be accessed by end users, combined with other data (Excel, CSV, SaaS sources etc.) and uploaded to the cloud without the need to directly access information locked up in corporate databases.
• How Indicee enable users to query data. For years Indicee founders listened to users wonder “Why can’t I just ask a question and get an answer?”. To address this, they built a dead-simple interface that allows users to ask business questions in plain English. Reports and visualizations are then produced on-demand from data uploaded to the cloud, eliminating hours of tedious cutting and pasting data from “canned“ reports into spreadsheets.
• How Indicee engages people.  Empowering and enabling business users in organizations to access data and information provides collaborative decision-making, transparency and alignment.  Indicee users can connect with each other and share data securely with colleagues or associates in an online community/social exchange.
• Indicee was founded by Mark Cunningham –16 years of entrepreneurial and technology experience previously Founder/CEO of Symmetrics, founder Crystal Decisions acquired by Seagate in 1996, and Fred Tummonds – 11 years of high tech industry experience previously at Business Objects where he was Vice President responsible for Enterprise Reporting.
• In early 2009 a beta version of Indicee software was released for commercial use, attracting both large and mid-sized businesses and partners including Mary Kay Cosmetics, Alco Ventures and Sage Software and others.

Check them out at http://indicee.com/

With the massive push toward cloud computing in the enterprise, there are some considerations that hardware vendors will have to come to terms with in the long run. Unlike the old infrastructure model with hardware bearing the brunt of fault tolerance, the new infrastructure model places all fault tolerance concerns within the software layer itself. I won’t say that this is a new concept as Google has been doing exactly this for a very long time (in IT time at least.) This accomplishes many things, but two particular benefits are that load balancing can now be more intelligent overall, and hardware can be reduced to the absolute most commodity parts available to cut cost.

Because the cloud does not need hardware to provide fault tolerance, the hardware required for a ‘cloud server’ is very basic. I like to think of these servers as netbook equivalents. Bargain bin motherboards, processors, RAM and hard drives can be thrown together to make a low cost commodity cloud server. A ‘Cloud OS’ and ‘Cloud FS’ handle the underpinnings as far as operating system and distributed file system. When combined in the right fashion, the Cloud Software Layer along with the underpinning Cloud OS and Cloud FS can literally allow one of these ‘cloud servers’ to be plugged in and auto-provision itself into the resource pool. When there is a failure of a component or an entire cloud server, the Cloud Software Layer can notify system administrators. Replacement is as simple as unplugging the bad server and plugging in a new one. The server will auto provision itself into the resource pool and it’s ready to go. Management and maintenance are simplified greatly.

Looking back at the hardware that will be used to make these cloud servers, last generation surplus parts are perfect for this type of implementation. Each individual server (or node in grid terms) has modest requirements similar to that of a netbook computer. The tasks that these servers will perform are well defined and it is the combination of hundreds or thousands of these servers that provide the real horse power behind the cloud. We see that netbooks can cost as little as $200 and I see no reason why these small cloud servers can not hit the $100 mark as they need no LCD display or peripheral ports, they can use cheaper standard 3.5″ hard drives and need no real casing to speak of (depending on the deployment method.) These units can even be racked in shelves of 4 units with direct DC power to each board. There would only need to be a single AC to DC inverter per rack with UPS to ensure power is flowing to the rack as a whole. The amount of heat being created will be far less than with a typical server, and it may even be possible to get the thermal thresholds down to the level where a bare heat-sink (without fans) can be used for the processors. This will also drastically reduce the amount of cooling needed in the data center. The possibilities are literally endless and it gets me excited just to think about this type of stuff.

Of course, all of this is dependent on the intelligence and robust fault tolerance built into the cloud software layer. As I said before, Google has already done this and has been using a similar infrastructure for a long time, so it is not a pipe dream. It is up to the individual hardware vendors such as Sun, HP and Dell to design and deliver a cloud server that will meed the needs of future cloud computing infrastructures. They will also need to deliver it at a cost that reflects the level of commodity the server now represents in the data center.

Oh, one more thing. I just wanted to note that it is not written anywhere that x86 has to be the processor architecture standard for this new breed of cloud servers. I can easily see a custom designed ARM processor fitting the bill.

Over the years, the heavy reliance on computers and servers has increased exponentially. With that reliance comes an increase in the overall number of machines in service. The server side of the business has ballooned recently to the point where businesses are faced with concerns such as electricity consumption, cooling and space constraints. Although the overall power (or capacity) of servers has grown rapidly over time, the efficient usage of these servers had not.

In most organizations, server utilization was in the low double digits. Very few businesses were able to utilize 80% or more of a server’s capacity, yet that server still consumed a lot of electricity and produced a lot of heat. Unfortunately, most applications are not designed to play nice with other applications and co-exist on the same server. This leaves quite an open door with regards to a solution. How can you cram more applications onto a single server thus raising the utilization? Furthermore, how can you do this and consume less electricity and produce less heat?

The answer is virtualization. The beauty of this solution is that you combat all of the above mentioned problems while adding some resiliency. Virtualization is the conversion of a physical machine to a virtual machine which is then placed on a host machine. Each host machine can contain several virtual machines. These virtual machines run isolated from each other, but share the physical resources of the host machine (such as RAM, CPU, Disk Space, Network Cards, etc.) This allows you to condense the number of servers (now virtual) that can run on a single physical machine. This will also have the side effect of raising the utilization of that host machine as more workloads are put on it.

An easy way to conceptualize virtualization is an office suite application. Years ago, it was not uncommon to have a typewriter for word processing, a ledger for tracking finances and an easel with a drawing pad for presentations. All of these took up a lot of space and they were not utilized all of the time. This is inefficient. Imagine if you could get all three to take up the same small space (footprint) and allow them to be used as needed without taking up more space. Along comes the office suite on a computer. Now you can do word processing, spreadsheets and presentations all within the same application on the same device. No additional space or resources are needed. This is an example of how virtualization works.

Take that approach to a web server, file server, DNS server, authentication server, mail server, etc. Those five applications typically required four or five servers depending on what operating system they ran and what applications were on them. With virtualization, we create separate virtual machines for each application, then host them all on one physical machine. If our mail server virtual machine crashes for any reason, it won’t affect the other virtual machines on the host. Each virtual machine is an isolated instance with it’s own operating system and application stack. Although all virtual machines on a host share the host’s resources, they cannot tresspass on the resources of another virtual machine.

So now we have five virtual machines running our critical applications on a single physical host machine. Wha happens if the mother board on the physical machine dies? Logically all of the machines on the host would die as well. This is where we extend the benefits of virtualization into a cluster of virtual hosts. Instead of one physical machine, we have two physical machines tied together in a cluster with shared storage connected to each. The virtual machines reside on the shared storage. If one physical machine dies, the control of the virtual machines on that host are passed to the second physical host. The second host can now power on the virtual machines that failed (since it also has control of the shared storage) and we’re back up and running. There is a lot more magic that goes on in the back end to make this happen, but I am just illustrating the functionality of the solution.

So in the end, we were able to reduce five servers down to two servers. In reality, we could place a lot more virtual machines on a physical host. Let’s say that we have three of each of the five original servers I talked about in order to service our entire business. That is fifteen servers in total. We would still use the above logic and place about half on one physical server, and the rest on the second. With this configuration, when one physical server fails, only half of the virtual machines go down and have to be spun up again on the remaining host. This ensures that our business continues as normal with minimal impact on the end user experience.

With fifteen servers condensed on to two physical servers, our electricity consumption has dropped to less than twenty percent of what it was. The heat produced by the servers has also dropped about as much, so the cooling requirement is eased as well. Lastly, we are using much less space to house two servers and a shared storage pool than fifteen servers. We also now need less servers, so less raw materials are used to make those servers. What do you get when you reduce electricity consumption, heat production, raw material usage and physical space (land)? You get a very green solution that is often referred to as Green IT or Green Computing. Within that space, there are vendors that further the ‘green level’ of the individual parts to bring you Green Storage, Green Servers, RoHS compliant machines, power management products and much more. The ability to green part of your business is increasing every day and one of the fastest and most stable ways to accomplish this is through virtualization.