InMage builds a village with HDS to deliver their CDP solution

The old saying, “it takes a village to raise a child” is true in the case of InMage, who has just signed an agreement with Hitachi Data Systems to offer their CDP solution as a HDS Program Product. InMage already has a number of customer wins under their belt, as well as an air of profitability to brag about, but they have yet to ring in the runaway success that their technology often times seems to warrant. The idea of being a one trick pony providing add on sales to the established major vendors has not proven as successful as the InMage team would like, so, it was time for a strategy change and partnering with HDS is just what they needed.

InMage, having been a player in the CDP space for over a decade, has awoken and realized they need an ecosystem “village” to provide complete solutions for their customers in today’s highly competitive environment. Moreover, partnering up is the only way some of the storage block bullies can build a complete solution with an ingredient as potent as InMage’s CDP. In this case, partnering enables both InMage and HDS to benefit. InMage gets access to the HDS sales force, services team, and installed customer base where they will undoubtedly be successful. Meanwhile, HDS previously only provided content replication within the same tier of HDS provided storage, so HDS gets a powerful heterogeneous replication and CDP tool to add to its arsenal. Now, in the area of CDP and replication, HDS has heavy weaponry matched to the competition with the likes of EMC with RecoverPoint, HPs with FalconStor-based CDP, IBM with SAN Volume Controller and Tivoli CDP for Files, and NetApp with Multistore and SnapMirror. HDS Professional Services team will undoubtedly use the InMage technology for migrations between other company’s storage and their own creating tiered storage solutions further increasing the HDS footprint within the data center. One of the more promising areas for the HDS team to focus on with the InMage technology is DR, delivering a HDS array into a secondary location thereby leveraging CDP to break the ties to the single vendor nature of most DR setups enabling end-users to benefit from leveraging multiple vendors to get the best solution.

With InMage being a HDS Program Product it enables HDS sales reps and channel partners to offer the InMage solution as though it was any other HDS software product, with the same terms and conditions, service and support.

Only time will tell if the child InMage has built a village for will best the storage block bullies. With the HDS partnership and prior wins in major verticals, InMage keeps going toe-to-toe, and it looks like they are winning their fair share of rounds.

Virtualization Energizes the Data Center Automation Segment

At Taneja Group, we’ve long focused on the impact of virtualization on the application lifecycle, from development through deployment and maintenance. The rise of virtualization has occurred in parallel with a renewed interest in data center automation (DCA), driven in equal parts by the codification of IT processes into the ITIL framework (providing a roadmap to automation), the budgetary and headcount pressures on IT departments, and the challenges of patching and updating virtualized apps and machines that are mobile, transient, and/or built on demand.

Vendor solutions for DCA typically emerge from one of three competencies: server and system management platforms, application development/test/release solutions, and application-specific lifecycle management tools, such as those for collaboration suites or relational databases. When it comes time to automate IT processes, we recommend taking stock of your existing trusted vendor relationships as well as your most glaring pain points: do you struggle to keep operating systems and servers patched to the right levels to meet compliance? Have you already automated the management of business-critical applications, and can the tools you’ve used be easily extended? Or, is a key issue the time it takes to configure and deploy applications developed in house – the apps-to-ops transition?

We’re closely tracking both established players and exciting new entrants in the DCA space. HP continues to enhance the DCA suite it acquired with Opsware (now the HP Business Service Automation suite), and IBM and BMC offer broad automation platforms for business service management as well. Platform Computing is extending its tools for managing resource-intensive, high-performance apps into the virtual server and cloud worlds, and Stratavia is on the same path, building on its strength managing relational databases from Oracle, MS, IBM and others. VMware has made a bold leap into the app lifecycle market by acquiring SpringSource, and rPath also comes to the automation game from the software development and release side, extending its packaging and delivery tools to also support the deployment and maintenance phases of the app lifecycle.

There’s no single best path to automation, but an incremental, iterative approach will typically be the easiest to sell, generate the least resistance among IT ops teams, and provide more immediate ROI data to justify further automation efforts. DCA has always been a market segment rife with bold claims. The general-purpose platforms often do little out of the box and demand heavy configuration and training, while the more focused solutions are generally not as easily extended to other automation challenges as their makers suggest. We’re watching closely to see who can prove out solutions with real-world customer success stories.

Enter Primary Storage Optimization

We have seen the onslaught of storage optimization technologies applied to secondary storage, i.e. backup/restore, archiving and disaster recovery, in the past six years. We call these Secondary Storage Optimization (SSO) technologies. The uptake of these technologies has been dramatic and we have seen paradigm-shifting influences in IT shops that have implemented these technologies. Today, backup processes are quicker and more reliable and restores from disk are almost guaranteed to work thanks to Secondary Storage Optimization technologies. IT is already keeping months of backups on disk and replicating deduplicated data to their DR site, using disk-to-disk replication. The role of tape is being dramatically re-defined in front of our eyes, but in the grand scheme of things we are still in the very early stages of data protection reform. It is hard to break thirty years of bad habits. Most IT shops, even the most advanced ones, are still 10-15% transformed in terms of using disk-based backup/restore solutions, especially with integrated data deduplication. Most are bringing additional application servers and data under the protection of disk-based solutions. The path is reasonably well defined by now and, in spite of the worldwide economic woes (or perhaps because of it), we see companies bringing more and more data under disk-based protection. As much as 40% of the companies that purchased these solutions in the past five years have added disk-to-disk replication already. That is a stunning statistic and speaks for the power of these solutions.

Enter Primary Storage Optimization. This is a relatively new area of endeavor for the industry. The thesis is “why not reduce storage requirements on primary storage, especially since it is the most expensive?” The idea has serious merit and all reduction benefits can theoretically be carried on to the secondary side, for a greater cumulative effect.

But wait just a minute. The rules for primary storage are very different. Applications are interacting with this data and application performance must not be impacted or IT would not buy it. This has led to the development of Primary Storage Optimization (PSO) technologies that work in a post-process method. That means data is shrunken in the background so application performance is not impacted. Some vendors have devised in-line products that can shrink data while it goes through the PSO appliance. Most work so far in the industry seems to be focused on file storage.

Technologies applied to PSO thus far have been unique, relative to SSO technologies. This was self-intuitive in that the problem being solved is different. Backups have a ton of duplicated data simply due to the way most backup software works. Extracting that duplication is different than extracting duplication from a primary data stream. At least so far we see vendors coming out with technologies that are developed from ground up to be for Primary Storage Optimization.

But even that is about to change. Several SSO vendors seem to have found a way to use the same technology in a different way to solve the PSO problem. While it is too early to call this, clearly there are benefits to having the same battle-tested technology applied to a new problem, assuming the efficiencies remain the same. We believe PSO is the next frontier in storage optimization. The activity under the covers in this area is extreme, just as it was six years ago for SSO technologies.

Our recommendation to IT is simple: As you implement SSO technologies and bring more and more data under disk protection keep an eye out for PSO technologies, especially those that will be synergistic with your SSO technology. The benefits of SSO are clear by now. We believe the benefits of PSO will be equally compelling. But it is the combination of the two that we are most excited about. Stay tuned!

Breaking Some Windows: VMware Acquires SpringSource

Starting with the vSphere launch earlier this year, VMware has steadily built out its message that virtualization is the new data center operating system. As the leader in virtualizing Windows applications, VMware is arguably the most significant threat to Windows dominance. They've isolated the OS into a virtual container, decoupled it from hardware, network and storage devices, and taken control of managing it (vCenter has become as indispensable a management tool as anything from Microsoft in many data centers) - in many ways VMware has Windows right where it wants it: walled off and compartmentalized.

Starting from this tactical advantage, VMware launched vSphere as a "Cloud OS," claiming it was the first general-purpose computing platform ready-built for mobility. Meanwhile, Microsoft countered by reinforcing its message that virtualization was simply a feature of the operating system (Windows Server 2008 R2, to be precise). For dedicated Windows shops, this message will likely resonate: if your key applications are already in Windows, and you've invested in Windows training and administrative tools, why toss it all to leverage some mobility or to do a little consolidation, especially if virtualization is essentially a freebie?

Now, however, to grab the reins again, VMware makes its largest acquisition to date, of the open-source powerhouse SpringSource. SpringSource isn't about the operating system: it's all about the application: developing, packaging, and deploying applications built for a framework, not for a particular operating system. It's the open source rallying cry - portability! reuse! flexibility! faster cycles! - tied to the "new OS" VMware message, and it's a savvy move. With virtualization moving rapidly out of dev and test labs, into production, and out to the Cloud, data center operators have turned their attention from virtual machine provisioning and lifecycle management to virtualized application performance and management. And, enterprise developers as well as VMware ecosystem vendors are delivering more and more applications as virtual appliances.

With it's AppSpeed product, acquired last year with start-up B-Hive and launched in June of this year, VMware got a first toe-hold in the virtualized application performance management space. With SpringSource, the company makes an even bolder move, acquiring a comprehensive open source application framework, complete with tools to build (Spring, Groovy & Grails), run (Tomcat-based tc Server, dm Java Server), and manage (Hyperic HQ) Java-based applications. VMware has made an investment in open source at least as important as Citrix's XenSource acquisition in 2007, and while Microsoft is still striving to deliver basic virtual server platform features to level the playing field at the virtual server level, VMware's moved on to the application.

With this acquisition, VMware hopes to take advantage of the considerable developer momentum behind the SpringSource open source tools and application framework. The battle for virtual infrastructure dominance is clearly moving up the stack, and VMware understands that in the end, both the utility and pervasiveness of the vSphere platform will depend heavily on the strategic value of the applications that run on it. We believe that with this announcement, VMware has taken another important step toward creating a development and run-time infrastructure that can span enterprise and cloud computing domains, while helping to bring the latter closer to enterprise readiness.

Application Delivery and Virtual Convergence

We've been discussing virtual convergence here on the blog recently, and the impact it has on IT operations. In particular, we've looked at the emergence of new types of administrators with cross-domain expertise and functionality. Virtualization owners with storage and server responsibilities, for example, or storage administrators who may take on a server role for virtualized workloads.

There are other communities in the mix, of course, and they're also facing new challenges. Development teams and application owners are rethinking their processes in a virtualized world as well. When the eventual target for an application might be a virtual machine from VMware, a physical server, or a cloud-hosted Xen-based farm, the development, testing, and validation process is more complex, requires new configuration and testing skills, and further strains IT support teams. We're rethinking infrastructure management - it's natural to rethink the application development, deployment, and lifecycle management process as well.

An interesting approach here is to rethink application packaging. Once an app is delivered to IT ops, the hard work usually begins: tweaking and tuning the server stack required to run it, including OS components, drivers, libraries, etc. Firms like rPath are building tools aimed at this challenge by developing a new packaging and release strategy that moves lifecycle management up the stack from the server to the application. With rPath, developers package up apps and bind them to the requires OS, data, and libraries they need, resulting in self-describing units that, theoretically, are ready to be deployed on nearly any virtual, physical, or cloud target.

Will this catch on? There are vested interests throughout the data center, and convincing infrastructure owners to give up some of that config and lifecycle control will take finesse. New release and source control systems can be a tough sell to the notoriously religious development community. But rPath and others are heading in the right direction. You can learn more from an actual client in this podcast interview with an rPath customer. Let us know what you think.

Podcast : Jeff and Dave discuss Virtual Infrastructure Optimization

Several months ago, Taneja Group introduced the concept of Virtual Infrastructure Optimization (VIO) as a category of solutions hitting the mainstream infrastructure management market today. This blog post is a summary of an attached podcast (click through on the title for the podcast), where two Taneja Group analysts (myself and Dave Bartoletti) dive into an overview of VIO. Note, you can find written articles discussing this technology online at InfoStor.com and at SearchServerVirtualization.com.

For many years, infrastructure management has dealt with a stack of services that are often sharply divided, and give the manager little opportunity to holistically view what is happening across the entire infrastructure. Even within single domains - for instance the network stack - we have multiple tiers of equipment and limited visibility across those tiers. Within the familiar OSI model, the enterprise may use certain types of equipment for physical layer connectivity, mainstream network equipment for routing and switching, and specialized intrusion prevention or over the wire analysis equipment for upper layer protocols. This example holds true in storage and often in server management as well. The administrator's ability to consolidate and view events and performance across these different domains has been little more than a pipe dream.

As the virtual infrastructure storms onto the scene, what has been a major stumbling block in the past for operations management is now becoming a crushing blow. The capabilities of the virtual infrastructure simply cannot be realized without visibility into what is happening, and what the impact of changes are. Management can't be executed, and automation can unleash unpredictable monsters. With no ability to determine the IO impact of a given application, nor the IO load on a given storage array, how can an administrator take action? Faced with a cloudy and opaque infrastructure, administrators dare not make use of capabilities like storage vMotion, and deny themselves the benefits of a fluid and adaptable infrastructure.

A number of vendors have for a long while worked toward revealing more information about the infrastructure, but their efforts often focus on enabling more effective planning. Planning isn't sufficient for day to day operational management, where action might be required to rebalance an out of balance infrastructure, or to free up infrastructure for a service event or outage. But a number of vendors are focused on analyzing on-going performance and/or over the wire events to give the administrator a view of what is happening in the moment, and then a big picture view of what has actually happened over time. Using these tools as fundamental instrumentation, the administrator can better fly their day to day infrastructure. We've termed this category virtual infrastructure optimization in part because the virtual infrastructure exacerbates the issues of limited visibility, in part because the virtual infrastructure makes reported data actionable and even automateable, and in part because the virtual infrastructure provides additional and better hooks on top of which active analysis and management vendors can build solutions.

The vendors on our short list most often look like Virtual Instruments, Akorri, Blue Stripe, and a handful of others. But many are nipping at the heels of the best in this category, or well-poised with the fundamentals required to eventually deliver a competitive solution. This includes vendors like Reflex Systems with an advanced security and IP networking pedigree, some array vendors like HP with their Performance Pack tools for EVA and XP arrays, and even the InfiniBand vendor Voltaire, as they become increasingly mainstream with multi-fabric solutions, but at the same time bring to bear their deep-visibility oriented Unified Fabric Manager.

Meet Taneja Group at VMWorld in San Francisco

Once again, Taneja Group will be all over VMworld Aug 31 to Sep 3 at the Moscone. If you're going, please let us know. We'd love to set up a time to meet you in person: just send a note to our intrepid schedulers, Nevin and Tara. We'll be running two Sessions this year. If you're registered, you can add them to your schedule now using the Session Builder: search on Taneja in the Speaker Company Name or VM3103 and DV2714 in the Session ID. We'll see you in San Francisco!

Tuesday, Sep 1, 11am-12pm: Session VM3103
How VMware Reduces Cost-per-Application and OpEx Costs

Speakers: Jeff Boles, Taneja Group; Alberto Farronato & Mark Chuang, VMware, Inc.

In today's economic environment, everyone is talking about saving money and lowering costs. It is generally understood that VMware solutions can lower capital expenses (CapEx), but how much does it really cost to deploy a VMware solution and what about savings on operational costs? And how does that compare to deploying other solutions? In a virtual environment, it's cost-per-application that really matters -- "How much does it cost for me to run my total set of business applications?" To calculate cost-per-application, one must factor in virtual machine density per host, hardware (servers, network, SAN), environmental (electricity, datacenter space), and administrative costs. This session will cover how to calculate cost-per-application and show how VMware solutions can deliver the lowest cost-per-application, even against so-called "free" solutions. It will also show how to quantify the operational / business savings from VMware capabilities like VMotion, Storage Motion, DRS, HA, FT, and SRM.

Tuesday, Sep 1, 2-3:30pm: Session DV2714
Optimizing VDI Storage with VMware View: Strategies for Success

Speakers: David Bartoletti & Jeff Byrne, Taneja Group

Virtual desktop infrastructures (VDIs) bring end users many potential benefits, including increased user mobility and flexibility, higher levels of security, and reduced administrative and maintenance costs. While the foundational computing and networking technologies are now in place to support virtual desktop infrastructures, the performance and provisioning of storage remains a major challenge to deploying an effective VDI. Traditional approaches to storage in a virtualized environment – whether networked or direct attached – tend to suffer from a number of problems, including low capacity utilization, poor boot performance and slow and complex provisioning. In this presentation Taneja Group dives into the four major storage challenges in VDI environments and offers proven approaches to overcoming them that utilize the latest developments in the View VDI suite from VMware as well as new array technologies from the leading storage vendors.

Virtualize with Confidence: New Tools for Performance Management

We've been discussing virtual environment performance optimization for a while, focusing primarily on the performance of the environment after it's been virtualized. But a quick scan of the virtualization administrator blogs will show you that, quite often, it's hard to get a handle on what constitutes good performance before an application is virtualized, or before a virtual desktop implementation is rolled out. And, of course, if you don't know in detail how things are performing today, or how much IT resource each app or user consumes - hourly, daily, on average, at end-of-month, etc. - you'll have a hard time proving you've maintained performance after a migration to virtual servers or desktops.

I've been watching this challenge closely, and two young vendors have shown me interesting tools lately. BlueStripe's FactFinder is an application performance management tool that installs quickly and monitors network connections to build a real-time map of application components and connections. You can drill down, hop by hop, to find bottlenecks or failures, but what I like best is the snapshot files it creates, which can be used to play back, Tivo-like, any period of time, in order to re-examine a performance problem. You can quickly benchmark the performance profile of an application, without much work, before you begin a P2V conversion. Once you're converted, run FactFinder against the new app and validate. The tool's not limited to monitoring virtualized application performance, like VMware's AppSpeed, so it's worth a look if you're still migrating critical applications.

New kid Liquidware Labs and its Stratusphere product monitors app performance in a similar way, adding its Connector ID technology to track connectivity to and from each virtual desktop in a VDI (virtual desktop infrastructure) environment. The company likens it to "caller ID" - you'll know down to the specific user who's using the most resources or doing something that's killing performance or response time. The maps in this case are specialized for VDI: Liquidware monitors performance for a while, then produces a report classifying desktops/users based on how suitable each is for conversion to a virtual desktop. They're working with leading VDI integration partners to lower the barriers to VDI adoption.

There's no need to virtualize in the dark: build a performance baseline without too much effort and you'll thank yourself later.

Moving Toward “Virtual Convergence” - Part 2

Last week we wrote about virtual convergence, and how it's leading to changes in traditional IT roles and responsibilities. This week we spoke with a half dozen IT managers at medium-to-large North American companies who told us how this is happening in their own organizations. All of these companies have appointed one or more virtual server administrators to manage their virtualization (mostly VMware) platforms. In one case the virtual server administrator has also assumed responsibility for physical servers and storage, demonstrating the importance of his role in the IT landscape. But in the majority of cases, the firms are relying on a senior IT infrastructure director of one flavor or another to oversee the convergence. Most of these IT executives have put in place new organizational structures and processes to plan the evolution of their virtual infrastructures and to more effectively manage the touch points between physical and virtual as well as server, storage and networking domains.

How are these new management structures working so far? Based on our sample of six, the results are mostly positive, though not without growing pains. For example, in one case, a virtual server manager in a particular department figured out a way to fund his own networked storage, rather than work through his storage counterparts in central IT. But this type of behavior seemed to be the exception. For the most part, IT practitioners are adapting to their new roles and decision-making processes. In what I considered the best example of this adaptation in a large IT organization, the senior IT manager formed ad hoc teams that are responsible for planning the growth of their virtual infrastructures as well as resolving conflicts. Each team includes at least one server, storage and network administrator along with a virtual server administrator. The multi-disciplinary teams bring broader perspectives to problem solving, and given their representative nature, ensure that decisions are more readily accepted and implemented. Another advantage is that the composition of the teams can be altered over time to reinvigorate group dynamics and to facilitate the sharing of a steady stream of new ideas and experiences.

As I talked to several of the IT managers, I felt inspired by their human spirit and creativity. I believe that the most successful IT organizations are embracing the changes driven by virtual convergence, and are figuring out how to capitalize on the best that the new technologies have to offer. More importantly, senior infrastructure managers are learning how to mobilize their teams to adapt to these changes and to bring out the best in their people. I didn't intend to write on this topic again this week, but I found some of these stories and experiences quite compelling.

In my next post, I plan to look at the impact of virtual convergence on management tools and approaches.

Dev/Test Lab Automation Expands Your Cloud Options

When virtualization first exploded on the scene, Data Center Automation (or DCA, or sometimes Run-Book Automation) enjoyed a renaissance within the vendor community. The push for IT process automation generally follows each major shift in technology. The story is usually, “you have to deal with new technology…why carry over labor-intensive and error-prone processes to manage it?” It’s a good story, and customers who plan for new processes to accompany disruptive technology implementations are shown over and over again to have much higher success rates.

This pattern is shaken up a bit when exploring how best to take advantage of off-premise or Cloud services, whether they are Infrastructure-, Platform-, or Application-as-a-Service (IaaS, PaaS, or SaaS) offerings. If you’re looking at hosted solutions, does IT process management or automation mean much? Isn’t that up to your hosting provider? Isn’t that what you’re paying for? The answer depends on the type of application or activity you plan to outsource.

Consider development and testing. Many dev/test teams work in a sandbox of physical or virtual servers and generally don’t get the level of IT support given to production applications. Dev/Test can often be a wild frontier: there’s a lot of freedom, but not much support. If you break something, you fix it yourself. If you don’t employ any sort of dev/test lab management processes (manual, scripted, automated, etc.) to control machine configurations, utilization or resource contention, you can’t expect any of these to improve when handed off to a service provider.

The good news is that, since virtualization often entered the enterprise through the dev/test lab, there’s a high level of experience and confidence in virtualization there, and lab automation solutions have had several years to grow up. Lab management tools also come in several flavors: in-house, fully-hosted, and hybrids – matching the multiple flavors of cloud available. You can start to automate your dev/test lab in house, with a private cloud (like IBM’s CloudBurst and others) as your infrastructure foundation. You could layer that with VMware’s Lab Manager or VMLogix LabManager to automate, or you could explore Surgient for a fully hosted solution. When you’re ready to move to a hybrid cloud, most vendors will be ready with cloud editions that enable you to move all or part of your infrastructure AND processes, as needed, off site.

Regardless of the path you choose, you should be finding the time to automate as much of your lab processes as possible, because you can’t move manual or ad hoc procedures anywhere. Lack of automation doesn’t just hurt the development cycle; it will soon start to hurt your ability to take advantage of the cloud.