Are We Looking at Convergence All Wrong? Part 3

Are We Looking at Convergence All Wrong? Part 3

Converged and Hyper-Converged Infrastructure. They could be opening up a new world that connects IT with the business, yet we’re focused on paying off technical debt.

Convergence is a trend that will disrupt how IT teams function. There is no one product that will solve your problems. However, by focusing on how to shift people, processes, and technology in the face of this disruptive trend, IT can become an asset to the business once again.

First, we talked about convergence as a trend. Last time, we talked about how converged infrastructure products are thinking too small.

This time, we’re going to talk about using convergence to connect information to the business.

Warning: this post will offer no pre-packaged solutions that will automagically solve your problems. IT organizations and companies like EMC have to collaborate to enable the changes in people’s skills and focus, organizational processes, and technology to get the real value out of convergence. Yes, I am no fun at parties.

Where Did We Go Wrong?

Converged infrastructure should enable businesses to connect their applications to their information without barriers, delays, or excessive cost. If you go to any business unit today, you’ll find innovative developers and business leaders stifled by their inability to get the resources necessary to implement their plans. Key data sits on IT resources, and the teams cannot get at their own data. IT cannot provide the appropriate services at a reasonable cost. As a result, they wall off access to the data, for fear that their teams will leverage the cloud and eliminate them. Of course, that’s exactly what their teams are trying to do right now.

Converging IT and the Business

IT must converge information access with business applications.

What value does IT bring to the table in a converged world? Information. IT controls most of the company’s information. By safely exposing that information to the business users, IT can move from “roadblock” to “enabler”.  This means enabling the business users to find the right version of the information they need, with the right service level, available on their compute platform of choice.

What does the business need? Support for their business applications – infrastructure and information.

Copy Data Management

To converge information access, IT infrastructure must solve the broad “Copy Data Management” challenge. This includes: distributed production access, collaborative access, operational recovery, disaster recovery, archive retrieval, and test & development. Moreover, access needs to be secure and efficient, so IT has to deliver: data scrubbing (e.g. cleaning out private information before sharing), optimized data movement (deduplication, compression, WAN acceleration, caching, etc.), search (by metadata or content), and heterogeneity (not tied to a given data format on-premises or in the cloud). It also has to apply to “legacy” as well as “new” information.

The Copy Data Management focus needs to be on business acceleration, not reducing cost. We have spent so long working on technical debt that we’ve almost forgotten what it’s like to enable growth. The first challenge, then, is changing the IT responses to business requests. These responses include: “Not unless we get massive new budget”” or “Unless you use an end-to-end IT solution, we can’t help you. (i.e. We can’t have you using cloud providers)” or “We need to organize a governance committee to evaluate this”. The second challenge is in finding technical solutions to solve the problem. “Short term” convergence delivers one-off solutions that optimize only a subset of the workflow, and data lakes often end up isolated from the rest of the infrastructure (and, yes, isolated from reality). In other words, IT leaders need to stop saying “no” as well as avoiding the pitfall of the “easy” technical solution that promises pain-free transformation.

IT’s Copy Data Management solution should: integrate into their existing environment, support new environments, and optimize flows for data, metadata, and management. I don’t believe there is a single product that can achieve these goals today (despite what anybody, including EMC, promises you). I also don’t believe that there are many IT organizations structured to deliver such a solution. Both products and IT organizations are built for silos. Therefore, the question for the IT team is – who do you trust as a technology partner to work with you on your journey to truly solve Copy Data Management?

Business Applications

To maximize the value of the information they’re managing, infrastructure teams must understand their business’s applications. Most companies do not know the relationship between different components of their application, much less how the application components connect to the infrastructure. A customer spent three months of broadcasting that he was decommissioning a server. Nobody objected. In fact, all parties asserted that it was a useless server. Of course, the day he shut the server down, the company website went offline! Another CIO spent three years classifying and understanding the interactions of his business applications, only to discover that the information was out-of-date before he could do anything with it. He then discovered he needed to find a new job.

If IT cannot understand the relationship between the business application and the production infrastructure, what hope do they have for the other copies? This lack of understanding leads companies to “replicate everything”, “archive everything forever”, or assert that subsets of the application are protected and pretend that it’s sufficient. Unfortunately, these approaches do not bring IT closer to the business; they just reinforce the wall between them.

Virtualization has made the sprawl worse. As “glue” VMs sprout up all over the place, one customer complained that they have not had a successful weekly DR test in 50 consecutive weeks. They never know what they need to replicate! Of course, at one point, we thought that virtualization might help IT gain insight into the business applications. The VMware vApp was going to provide the information that connected different parts of an application, so we could understand the business application. This has not come to fruition. Instead, server virtualization has led to more entropy between the business application and IT infrastructure.

The next inflection point is upon us: containerized, micro-service based applications. Regardless of your preferred containers and container management systems, applications and data are about to become even more dispersed. On the other hand, as these tools roll out, we have another opportunity to work with the business and technology tools to leave enough bread crumbs for IT to understand the components that comprise an application. The time, however, is now, before “just do it” processes become de facto standards.

By beginning the journey to understand the business applications now, IT will be prepared to handle containerized, virtualized, hybrid-cloud applications. As with Copy Data Management, there is no tool that will solve the challenge of mapping business applications to the infrastructure. This, too, will be more of a people and process change than a technical change (disruptive technical change always disrupts people and process more than anything). However, by finding the right technology partner, IT organizations can begin to better understand their business.

Conclusion

More than any other technical trend, convergence can reshape IT. Convergence will either drive businesses completely to the cloud, where they’ll trade service levels for agility, or it will reinvigorate on-premises IT for the next decade (where they’ll leverage cloud as an asset). Incremental steps to reduce technical debt are fine, but IT will only survive and thrive if it can become a business enabler, rather than a cost center.

IT’s core value comes from its management of the company’s information. The business wants that information available to its applications. By changing their people’s skill sets and focus, processes, and technology, IT can deliver Copy Data Management that drives the business applications – wherever they run. That journey has to start now.

Next time, we’ll talk about how convergence is more than just packaging – it can change the way people consume IT assets.

IT must converge information access with business applications.

Stephen Manley @makitadremel

Are We Looking at Convergence All Wrong? Part 1

Are We Looking at Convergence All Wrong? Part 1

Converged and Hyper-Converged Infrastructure. Never has the IT world generated more hype and madness about any topic (other than flash, object, disk-based backup, CDP, appliances…). As I spend more time with customers, engineers, and analysts, I’m convinced that we’re looking at this myopically. Convergence, has the chance to be a true technology disruption (yes, I mean a REAL disruption like CLOUD, not just the “Hey we’re using a new type of storage media or processor” disruption-of-the-day). But only if we start looking at it the right way.

There are three areas in which we’re focusing on the wrong things:

  • Convergence as a Product or Architecture. Convergence is an underlying trend, and there will be many products and architectures that lay claim to the title “Converged Infrastructure”. By constantly looking for the “right” product or architecture, most organizations will miss the point.
  • Convergence at a small scale. Most companies focus on convergence with a small ‘c’ – converging production VMs, converging backup software and hardware, or converging the stack for a single application. Convergence will drive much more fundamental changes.
  • Convergence as a packaging of IT infrastructure. While packaging is beneficial, it is far less important than how convergence can drive the new consumption models.

In other words, just as happened with cloud, most of us are not thinking broadly enough when it comes to Converged Infrastructure.

Convergence – It’s a Trend, not an Architecture or Product

When it comes to significant, abstract IT trends, people need to make them tangible. Engineers need to build and test something. Sales teams need to sell something. Customers need to buy and deploy something. Every time I talk about vision, strategy, or solutions, I can see the audience itching for something concrete – product.

Therefore, when talking about Converged Infrastructure, people want to focus on product. Everybody feels more grounded when talking about Vblock, FlexPod, Exadata, Nutanix, or VxRail because they can talk about price, features, and performance. They’re also relatively comfortable talking about Converged vs. Hyper-Converged architectures because they can compare and contrast different architectures via existing products.

Convergence, though, is more of a trend than a product or architecture. Convergence is the trend of eliminating custom work by buying pre-bundled solutions to help IT become more responsive and cost-efficient. When layers of the IT stack commoditize, the value of manually integrating “best of breed” products diminishes rapidly. IT organizations are better served buying pre-bundled solutions and focusing their time and energy on parts of the stack that drive higher business value.

The convergence “trend” happens across all industries, including IT infrastructure.* Most organizations don’t build their own servers anymore because the components became commodity. Most companies don’t build their own NAS systems (e.g. by buying SAN storage, host-based volume managers, and host-based file systems) because the components are commoditized and the differentiation comes from the file system. Security and networking appliances continue to converge as core components commoditize. As flash makes I/O performance easier to deliver, storage is going to become more of a commodity – making it ripe for convergence. The trends for hypervisors and backup make them appear to be on the path to commoditization, as well.

*Of course, in response to core technical trend changes, the “convergence” trend can be replaced by a “divergence” trend – as we see some large companies build their own servers, file storage, etc.

Convergence is a trend, not a product or architecture. There will be many incremental products that will converge components of IT infrastructure, and their incremental value will make them worthwhile. There is no “one right answer” today, however, because the convergence wave will inexorably sweep away most of these interim products and architectures. The value in convergence will come from understanding what business goal IT wants to deliver and what parts of the IT stack will be commoditized and converged.

Over the next two posts we’ll explore the optimal targets for Converged Infrastructure and why the discussion about packaging is just the beginning of the discussion about how we buy and deploy IT infrastructure.

Converged and Hyper-Converged Infrastructure products are over-hyped right now.

Convergence, as a truly disruptive trend, is currently being underestimated.

Stephen Manley @makitadremel

The Summer of Love And The Urge to Converge

The Summer of Love And The Urge to Converge

Before you get too excited, this is not some tawdry summer romance saga of data center passion and desire.  There’s no need to alert the folks in HR.  But now that I have your attention, what could the Summer of Love of 1967 and today’s shift to Converged Infrastructure possibly have in common?

One commonality is they both represent movements by groups with views that were considered different than those of the “establishment”. They often challenged the status quo and were united by the desire to make to make things better.  And as their views and ideas began to drive meaningful changes, they became accepted.

I admit the analogy is a bit of a stretch. But another key point they share is that for any major movement to take place, there needs to be a coalition of the willing to drive transformation. There is also often a tipping point that provides the catalyst for moving these transformations into the mainstream.

For the “hippies” of the 60’s, the tipping point happened when they converged at the Junction of Haight and Ashbury in San Francisco to provide the catalyst for the “Summer of Love”.  I’m not saying 10,000 IT members are about to converge on the VCE HQ in Marlboro MA (although that would be very cool and they do have a very nice EBC J). But there is a clear trend of increased interest, with more and more user discussions revolving around Converged Infrastructure. One could argue that the tipping point is very close, if not happing right now.

The latest numbers certainly support this movement. IDC estimates that Converged Infrastructure will represent a $3 Billion market by 2018, with a growth rate at over 13%, one of the fastest growing segments in IT infrastructure.  The reasons for why IT is moving to CI certainly support these growth rates. According to IDC; 4X faster time to deploy, 20X better availability, and 2X more productivity for IT teams.

As many folks know, Converged Infrastructure makes it simplifies technology deployment with standardized, prepackaged blocks. These blocks also have API hooks that make them easy to integrate into a range of applications. Provisioning infrastructure resources is an automated process with a simple workflow. CI provides for efficient consolidation with the ability to throttle resources to support different workload types. The throttle controls, however, have typically been performed by dialing in CPU cores, amount of memory, and the number of storage GB’s needed by the app. In the world of servers, provisioning and tuning of CPU and memory resources is a highly abstracted and simple process.  Select the number of cores and amount of memory, and away you go.

Provisioning and tuning storage for different workloads, however, can be a fairly manual and cumbersome process to plan, set up, and administer (i.e. managing different FAST policies for specific apps to optimize storage performance). As a result, many admins simplify their approaches and create a single bucket of “one size fits all” storage (i.e. having a single FAST policy and providing apps a single, similar level of performance). This does not mean different policies or multiple configurations can’t be used to control service levels of the storage. But for most admins there is a tradeoff between keeping things as simple as possible versus having a more having a more granular control.  Most choose simplicity.

The tipping point for Converged Infrastructure happening today is that the need to make this tradeoff is disappearing.  The ability to provision storage by Service Level Objective removes the complexity of managing different array policies or manually setting up specialized storage configurations. The optimization of storage services for different workload types can now be both simple and predictable.  With SLO, it’s possible to efficiently consolidate storage across hundreds to thousands of apps with different workloads. Each app can be individually managed based on its compliance to its service level. New apps can also be easily on-boarded, and SLO’s for existing apps can easily be adjusted with a few simple clicks.

SLO management also provides a critical workload planning service that understands the current load running across the storage.  Admins can monitor the amount of headroom available and the system will automatically calculate the amount of additional storage resources and performance needed to support the SLO for new apps.  The end result is a level of predictability that makes it possible to confidently support a range of workload types, and addresses what we call the “half a cookie conundrum” faced by many IT admins.

If you have young kids, you will understand what I mean. If you give your kid a half of cookie, generally they are happy to get a nice treat.  But if you give your kid a whole cookie, and then take half of that cookie back, you most likely will have an unhappy kid you need to deal with to justify why you took away half their cookie.

It’s a similar situation to what an IT admin goes through with their users and app owners.  The first sets of apps ran great because they had all of the resources to themselves, i.e. the got the whole cookie. But as apps were added, they had to share the infrastructure, and in their minds, they had to give up some of the cookie they were first given. Provisioning via SLO allows every app to get a consistent, predictable level of service, whether they are the first app to be added, or the last app that fits. It provides a critical capability and solves a real problem, and provides the tipping point for many large scale CI deployments and consolidations.

The concept is certainly gaining traction. The roll out of the capability is resonating with IT admins, and feedback has been highly positive. As the adoption continues to increase, it’s easy to see we are quickly approaching the tipping to the transformation of Converged Infrastructure becoming the deployment option of choice for an even greater range of applications and workload types.

The American psychologist Timothy Leary summed up the Summer of Love by defining the movement as a generation who joined together to “turn on, tune in, drop out“.  He later defined what that statement meant as; “Turn on” to activate your neural and genetic equipment, “Tune in” to interact with the world around you, and “Drop out” as an active, selective, graceful process of detachment.

The statement might apply to today’s movement to Converged Infrastructure as “Turn on” to activate your urge to converge, “Tune in” to interact with the workloads around you, and “Drop out” as an active, selective, graceful process of detachment from complexity.

See you on the bus to Marlboro.  Peace.

Scott Delandy @scottdelandy