IDEAS Insights

Rise of Virtualization Could Boost Grid Computing

The growing popularity of virtualization may renew interest in Grid computing, as users couple virtual machines with job scheduling software and take advantage of virtual infrastructure to share computing resources on a global basis. One of the key benefits of virtual machines is that they dramatically simplify the process of migrating workloads across a network, a process that is a fundamental underpinning of Grid computing. Because virtual machines isolate applications from the details about the servers on which they are hosted, it becomes possible to move virtualized workloads from one machine to another with a minimum of disruption. Many virtualization platforms enable the entire state of a running virtual machine to be captured in standard files, which can be seamlessly transported across a network using shared storage. With the ability to coalesce an entire server into a few files, it is relatively easy to move them to another host on the fly, and with live migration support, virtual machines can be migrated while their applications continue running.

The vision of tapping compute power from a "Grid", i.e. a global set of resources that can be tapped into any time, and from anywhere, continues to appeal. Grid computing remains a powerful technique for enabling the rapid completion of very compute-intensive applications, and also promises some economic benefits, such as utilizing spare resources in other time zones that are in a period of reduced business activity, or -- given the growing concern in the IT industry with constraints in natural resources -- taking advantage of datacenters where power and space are more affordable. The opportunities for Grid computing to boost collaboration and reduce costs generated some excitement in the industry a few years ago, and prompted some systems vendors to invest considerable resources in developing and marketing Grid solutions.

However, much of the Grid computing vision yet remains to be fully implemented, except with relatively specialized applications, and in certain types of environments, such as research and education organizations, or leading-edge financial services companies. One of the barriers to the adoption of Grid computing is the relative complexity of customizing and adapting workloads to make them suitable for hosting on a Grid. Job scheduling software, which automates the assignment of workloads to hosts based on the priority of a job, is relatively mature, but it has traditionally been used to manage resources at the level of individual applications. In order to ensure proper execution of an application that is submitted to a Grid, it is necessary to provide the application with the software environment that it needs to execute. A Grid application may depend on details about the OS platform on which it runs, as well as tools and middleware needed to support its functionality.

Virtualization makes packaging applications for deployment on a Grid much more practical than before. After a workload is packaged in a virtual machine (or virtual appliance), it can run on any host supporting that VM format, and any software needed by the application can be included in the VM without disrupting the host on which it runs. Indeed, bonds are starting to emerge between VM platforms and Grid scheduling frameworks, whereby the Grid schedulers dynamically assign resources at the level of hypervisors, rather than applications. For example, XenSource recently announced a partnership with Platform Computing, one of the leading suppliers of Grid scheduling software. Platform already began supporting VMware in 2005, and it will now integrate and bundle XenEnterprise v4 with the latest version of its VM Orchestrator (VMO) package, which means that Platform's software will be able to dynamically allocate shared resources to the Xen hypervisor based on the priority of workloads it is hosting. Several VM vendors are growing their own set of distributed resource management tools optimized for virtual machines, including VMware with its Distributed Resource Scheduler (DRS), and Novell with its ZENworks Orchestrator, but tools such as Platform's have already been proven in use on a global scale. 

Another example of the convergence between VMs and Grids is the rDistributor project, a collaboration between rPath, the developer of a virtual appliance platform, and the Open Science Grid, a consortium of research organizations. The project takes advantage of rPath's Conary distributed software management system, which makes sure Grid-based applications have the necessary software to execute by automatically installing the components they need in their VMs. Without virtualization, such automatic installation of components would be impractical, since it is unlikely remote hosts would allow arbitrary installation of support software for purposes of running a temporary application.

Virtual machines are now being deployed in organizations of all sizes, and with many types of workloads. Most users have yet to begin expanding the scope of their virtualization efforts from single servers to multiple systems. But over time, users will become increasingly comfortable migrating VMs across the network, in pursuit of reduced downtime and greater responsiveness to changing workloads. As organizations build out virtual infrastructure on an ever larger scale, many of the benefits that have long been promised for Grid computing will finally reach the broader market through the path of virtualization.

Intel Shows Off at the 2007 IDF

The 2007 Intel Developer Forum was held last week in San Francisco. This year, Intel changed things around compared to past years. After spending time deciding which sessions I should attend and setting up a schedule, Intel Analyst Relations announced that our schedules had been superseded by a special track for press and analysts. The reason undoubtedly was to ensure that a unique set of specially tuned messages went out to each group of attendees. One analyst jokingly remarked that the new track was designed to keep the good material away from the press and analysts. To ensure that you followed the right track, Intel included an RFID tracking chip on the badges of general attendees and industry analysts. For some reason, the press was spared the honor of an RFID chip.

The conference exhibited the typical Intel "wow" factor as speaker after speaker demonstrated exciting new technology and R&D concepts in a grand fashion. Intel CEO Paul Otellini set the stage for all three days with a keynote titled "Extreme to Mainstream." The overall atmosphere of IDF was similar to a rock concert with blaring music and bright lights focused on the stage. As each speaker was introduced, he or she came out from behind the curtain and simultaneously appeared on three 50-foot video screens facing the audience of 5000. One of the highlights of this year’s IDF was the rare appearance of Dr. Gordon Moore, the father of "Moore’s Law." Dr. Moore was interviewed by Dr. Moira Gunn of NPR and both spent time reminiscing about important past events in Moore’s professional life. Intel CTO Justin Rattner was especially entertaining – appearing as his virtual world alter ego from the website "Second Life." The virtual Justin is a young, blonde, California surfer dude who has six-pack abs and a barrel chest – physical characteristics that differ sharply from those of the real Dr. Rattner. More than a few chuckles could be heard from the audience as the virtual Justin appeared on the screens.

Aside from the entertainment factor, the role of the IDF is to highlight Intel’s capabilities and to focus the audience on current and future products. In the server space, new information was relatively hard to come by because Intel revealed most of it prior to AMD’s launch of the quad-core Barcelona and Intel’s launch of Tigerton. To ensure that AMD never pulls ahead of Intel again, Pat Gelsinger described Intel’s new "Tick Tock" model for processor development. With Tick Tock, the "Tick" is a shrink/derivative of the previous generation processor, followed a year later by the "Tock," which is a new microarchitecture based on the existing process. This new R&D strategy not only allows Intel to be very predictable, but it also saves valuable R&D dollars by creating a common architecture across desktop, server, and mobile processors.

In the server processor space, the 45-nm "Penryn" and the follow-on "Nehalem" took the spotlight. Penryn is the "Tick" that follows the quad-core 65-nm "Tigerton" processor that was just launched. Penryn is in production today and Intel expects to launch it on November 12 for high-end servers and desktops. Penryn shares the same Core Architecture with its predecessor. On the "Tock" side, Intel previewed "Nehalem," which features a new, modular design, allowing Intel to create semi-custom chips for many market segments. Pat Gelsinger was very clear that Intel was not announcing a new microprocessor marketing or development strategy, but rather the ability to address different market segments if Intel chooses. Nehalem will feature eight cores on a single die and each core will have two threads. With Nehalem, Intel is integrating the memory controller and introducing "Quick Path Interconnect." The latter feature is similar to AMD’s HyperTransport in that it allows multiple chips to be easily connected together to form a large SMP server. Nehalem is also the generation where Xeon and Itanium will finally become socket compatible. Nehalem will be followed by "Westmere" and "Sandy Bridge," both of which are based on Intel’s 32-nm process.

Despite this year’s IDF being a little short on server content, not all was lost. Intel demonstrated that it will be solidly in the lead over AMD with its new Tick Tock microprocessor development strategy. AMD simply does not have the vast R&D resources needed to consistently match Intel’s aggressive schedule. Intel even feels AMD may have trouble moving to its next generation 45-nm process, based on the technical issues Intel encountered moving to its own 45-nm process. With 45-nm, the traditional silicon dioxide dielectric layer was too thin to function properly and Intel had to develop the hafnium high-k metal gate technology to take its place. With AMD finally under control in the server space, Intel has the time and resources to focus on other areas, such as mobile devices and the entertainment industry. Maybe Justin Rattner’s highlighting of Second Life is a harbinger of Intel’s future. Even corporations need a second life at some point. Either that or Justin is demonstrating he has a little too much time on his hands these days.

What Features Really Make a Product Green?

IDEAS recently attended HP’s 2007 Asia Pacific & Japan Industry Analyst Summit. This is an annual event for industry analysts around the region where HP provides updates on its business and showcases strategy for the coming year. Not surprisingly, ‘Green’ was a strong theme.

HP even flew in its Vice President, Environment, Pat Tiernan, who hosted a 2-hour session on the subject. This was a significant amount of time to devote to a subject for a two day event that aimed to cover all aspects of HP’s business units.

(Pat Tiernan, Vice President, Environment, presenting at the HP Asia Pacific & Japan Industry Analyst Summit 2007)

Mr. Tiernan presented some high-level findings from research that HP has commissioned in the Asia Pacific region that found “44% of respondents are willing to pay a premium” for ‘Green’ products.  The research also found that the likelihood to pay a premium is highest for products that are energy efficient.

Although Green encompasses many aspects, such as product re-use and recycling and use of materials etc., energy efficiency is a key focus. This is where clients are most willing to pay a premium up front, because this investment is directly linked with some sort of tangible economic ‘dividend’ in the medium to long term. Basically, pay more up front for products that will give back the value of the premium or more, in lower energy bills over time.

When it comes to power and energy issues they apply all the way from components, to products, management practices and how you use individual products or groups of products together (virtualization being a key example), right up to control of the IT climate, or the datacenter.

Much of the focus of the vendors to date has been at the higher levels (management practices and datacenter) where work can be done whilst products catch up in terms of tangible energy efficient features. The power and cooling aspects of Green are a relatively new phenomenon and product development cycles for all companies are now working to catch up with the ‘Green theme’ of 2007.

At the Summit IDEAS was keen to drill down with HP executives to better understand what product features HP has, or is working on, that will contribute to this ‘Green’, and particularly ‘energy’ dividend.

Speaking with Tony Parkinson, Vice President and General Manager, Industry Standard Servers, he discussed some of the things to look for. This is not an exhaustive list by any means, but gives a flavor as to what features, besides a product’s ability to virtualize, would make it ‘Green’, especially from an energy perspective.

• The first is of course to use newer generation components, such as new generation multi-core processors that use less power and/or power per unit of performance. Intel’s recent release of its Quad-Core Xeon 7300 series processor has a Thermal Design Power (TDP) rating of 50W, contrasting with previous generation single core processors of the recent past that have a TDP of 120W. Disk drive technologies, such as the new generation 2.5” SAS drives can draw less power than previous generation technologies.
• Power supplies are another area of focus. Power supplies are most energy efficient at higher levels of utilization, so a lot of work is being done in the industry to find ways to build more efficient power supplies and ways to better utilise them to maximise their efficiency.
• Better cooling capabilities of products that generate heat is also an area of focus. If a product’s cooling capability can be designed in such a way as to allow it to run in higher ambient temperatures, then that in turn puts less pressure on the datacenter cooling requirements.
• A less obvious area of energy product design is one of putting ‘awareness’ into products. A product needs to be able to be aware of its environment, in order to be able to react to changes that may occur. Products will begin to appear with in-built sensors to monitor the internal system temperatures and/or electricity loading.
• Then of course there is the management layer (software) that will monitor and drive any responses that needs to be made in reacting to any system environmental changes.

These are just a few of the features to look for in products over the coming months and years, as products designed for this new age of energy efficient computing start to see the light of day.

Storage Is a Key Aspect of Server Virtualization, Part 2

To illustrate the relevance of storage to server virtualization environments, last week’s enterprise storage post discussed some of the significant storage announcements that occurred at VMworld. Of course, the companies mentioned in the previous post were hardly all inclusive; several other storage companies announced storage products targeted for VMware. The following continues where last week’s post left off, highlighting the announcements that we believe will have the most significant impact. (For a complete list of product announcements, please check the VMware website).

Storage virtualization provider DataCore announced on September 10 that its storage virtualization and SAN management products now support virtualized server platforms or virtual machines (VMs). While most vendors at VMworld were understandably rushing to support VMware, DataCore announced support for server virtualization platforms Xensource and Microsoft Virtual Iron in addition to VMware. DataCore’s SANsymphony, Traveller, and SANmelody products allow virtual SAN storage servers to be created on VMs. Such a capability provides an extra level of integration and collaboration between server and storage virtualization. DataCore’s press release states, and rightly so, that this announcement is significant in that continues to move storage virtualization more toward hardware-independent software solutions, as opposed to single-purpose hardware devices. Although several virtualization solutions today are hardware or device based, as the industry matures, we’ll see more hardware independent software options become available as these products evolve.
 
Other notable announcements were:

• Acronis’s True Image Echo appliance has been expanded to back up and restore from virtual servers in addition to physical servers. Acronis supports multiple VM vendors: Microsoft, Xensource, and Parallels.
• EqualLogic will integrate its PS Series with VMware’s Site Recovery Manager for automated replication and recovery.
• Pillar Data Systems has joined VMware’s Technology Alliance Partner Program.
• Change management company Onaro will add VM Insight to its SANscreen software suite. VM Insight will track and manage changes to VMs.
• QLogic (competitor to Emulex) demonstrated that its virtual HBAs allow live data migrations, i.e no requirement to take down and reboot servers.
• StorServer will offer an agent with management, scheduling, and reporting for backups initiated by VMware Consolidated Backup.

These types of announcements demonstrate how server virtualization and storage virtualization are becoming more integrated. As virtual environments grow, the demands on the storage infrastructure will increase dramatically. Virtualized storage environments – that are fully interoperable with VMs – will be required to meet the increasing demands on performance, availability, and energy efficiency. 

Users Committed to Virtualization Offered Broad Array of Solutions at VMworld

One of the most impressive experiences at VMworld this past week was a tour of the nearly 150 exhibitor and sponsor exhibits that showcased extraordinarily diverse solutions for managing virtualized resources, and applying virtualization in many other parts of IT infrastructure. Virtualization of I/O resources such as storage and networking are clearly on the upswing. Based on user reactions, there also appeared to be particularly heavy interest in virtualizing desktop systems by means of Virtual Desktop Infrastructure (VDI), whereby standard desktop operating systems are hosted in virtual machines that users interact with remotely by means of thin client devices.

As organizations increasingly deploy production server workloads on virtual machines, there is a growing need for more sophisticated tools to manage performance in virtual infrastructures -- and several vendors were on hand showcasing products designed to ease the management burden, focusing particularly on maintaining performance. Finally, several vendors were addressing the growing user concerns about security issues that can arise in virtualized infrastructures. While much of the early attention around virtualization focused on its value as a tactical tool for consolidation, test/development, and maintaining support for legacy workloads, organizations now appear to be driving virtualization into many other critical parts of their infrastructure. The sense at VMworld was that an entire industry is emerging that is retooled around virtual infrastructure.