Egenera Blog

Being here in our Nation’s Capital at this time of year, we’re on the verge of dramatic changes.  I’m not talking about just the obvious changes coming November 4, but to Federal computing environments and their data centers as well. No question about it - change is coming at us from every angle.

The list of considerations for change is long…perhaps most notably for Federal data centers, space is limited, power consumption and cooling requirements are on the rise, and complexity has never been higher or more of an impediment.  These events are moving most Federal agencies closer and closer to virtualization.  This isn’t news in and of itself, but it’s still noteworthy given that this is a sure sign that virtualization has gone past the point of “novel.”  Most analysts agree that we’re now well into the next phase of virtualization deployment.

In a recent article in Government Computer News (GCN), a Federal survey polling nearly 300 executives saw most deploying server virtualization, and 33 percent implementing hypervisor technologies specifically to save energy.

However, according to the survey conducted by Actionable Research, respondents said the technology was far from perfect.  Of those using virtualization, 24 percent have experienced a disappearance of virtual servers from their systems, and 18 percent reported having permanently lost a virtual server. Additionally, 45 percent of respondents said they had concerns about the lack of expertise that their IT staffers had with virtualization, and 44 percent said they were concerned that virtual servers could fail because of a component failure in a physical server.

These aren’t small issues…we’re talking significant issues and percentages with significant failure costs associated with them. Add in the concerns about lack of experience - and the growing complexity of the data center - and it’s a whole new world out there, so to speak.

This is where change in the data center isn’t a “nice to do” anymore - it’s mandatory.  Solutions like our own PAN Manager software become even more valuable in times like these.  Federal - and really all - customers need to be able to create, manage, and monitor both physical and virtual environments easily and affordably.  Solutions where N+1failover and disaster recovery aren’t just marketing catchphrases, but repeatable, verifiable and reliable approaches to guarantee up-time no matter what happens have never been more important.

I suspect these are changes we can all get behind. What’s on your list of changes to watch?

There are a number of challenges facing data center managers today – not just the sudden curtailment of investments due to the economic situation.  Data centers are in the midst of one of the largest transformations in decades.  The shift is driven by more than just new technology.  New business models that extend corporate systems to include partners and suppliers; the redefinition of traditional IT organization models and processes; social computing; and new expectations of how IT will support corporate decision making are just a few of the disruptors.  Boil that all down and what you’ve got is a data center manager constantly managing not enough floor space, skyrocketing electricity bills and heat, rampant VM sprawl and pressures to contain costs while improving customer service levels.  While most data centers are juggling all this, I think we can agree that this dog just doesn’t hunt. It’s neither sustainable nor scalable.

The move to a reliable dynamic data center is a way to a more sustainable model. Notice I didn’t say vision - it’s much more real than slideware.  So what is a reliable dynamic data center? I like the definition set forth by the Burton Group.   Together with one our OEM partners, we’re defining the building blocks of a reliable dynamic data center.

I consider the reliable dynamic data center a model that leverages the adoption of virtualization that’s already happened (server virtualization).  Because without it, companies can’t achieve the agility, flexibility and reliability needed to evolve. And not evolving isn’t an option, right?

Server virtualization was the building block.  Cost reduction is what drove that step into mainstream adoption. The side benefit of the adoption of hypervisors and other forms of virtualization was that it allowed companies to move on to the second building block – improving quality of service.  Advanced adopters of virtualization realized that it positively impacted their ability to meet SLAs along with making it easier to automate provisioning, high availability and disaster recovery.  The third building block is improving agility, speed, efficiency and optimization through application workload consolidation. By automating dynamic allocation and balancing computing resources via automated business policies along with unified fabrics, standardized management interfaces, orchestration engines and IT governance, companies are firmly on the path to achieving a reliable dynamic data center.

Unifying the management of physical and virtual infrastructure today is one cornerstone of the ‘data center of the future’.

What are your thoughts? What are your building blocks?

Tags: Dynamic Data Center

Server Virtualization started simple enough - a way to consolidate multiple application servers to a single physical host. We might say it was learning to walk before running. At first we saw a move of test and development applications making the move to reduce costs and footprint. Over the last few years however we’ve entered the running phase with increasingly more production systems moving onto hypervisors like VMware and Xen. However for the most part this has been applications and functions that are generally considered to be on the “edge.”

Edge servers are generally considered to be those applications or functions that are “simple” and not business and/or mission critical. This could be anything from file and print servers to simple applications serving departmental functions. Applications that will, in general, not cause any serious impact to the business one way or the other.

So the line of thought has always been that these servers do not need high availability or disaster recovery. This was true! However as true as it may be that only minimal impact may be felt by the business if one edge application goes down, what happens when you have 20-30 edge applications on a single server via a hypervisor? It is kind of like an old saying “a single sting from a bee hurts, but a thousand can kill you.” So even if it was true in the past that you could lose a server, now wither a server running possibly 30 edge servers, that can, and more than likely will, affect the business.

Most businesses have been so busy over the last few years consolidating their data centers that they have moved from what were simple edge servers to a consolidated environment. This effectively moved them into a business critical space. This isn’t a bad thing, but it needs to be planned for and for the most part, it’s been overlooked.

With more and more servers moving onto consolidated platforms, the more we’ll see a need to move to a more dynamic data center model overall. It’s the natural evolution of virtualization after all.  The edge is disappearing right before our eyes.

My colleague, Christine Crandell, did a great post on her initial thoughts on how the financial crisis will affect all of us - customers, vendors and individuals. And importantly, that the opportunity for impact starts with our attitudes. It got me thinking…about who really “wins” in times like these. Value is more important than ever - but it’s more than dollars and cents.

So, IT budgets are once again under the microscope after recovering somewhat over the past few years following the post dot-com/Y2K/ERP binge of the late 90s, so warns the Burton Group (and so many others).  Inevitably, it is the vendors that can demonstrate true customer returns and value-add that will emerge the strongest through the period.

The products that deliver the highest customer ROI are based on simplification and integrated delivery.  The best products eliminate clutter, not add to it.  And as much as is feasible, the product should contain all the functionality necessary to complete key tasks while providing robust mechanisms to snap into other components at the functional boundaries. Make sense?

Often there is a temptation to simply add software layers to fill in some functional gaps and to create super views to mask complexities below.  But many of these approaches result in systems that very complex and require highly specialized training, significant set-up effort and continuous maintenance.  These costs hugely detract from the benefits.  Worse, costs grow disproportionately with scale, so the largest environments face the greatest strains.  During good times, these escalating operational costs may not be as noticeable.  During downturns, they stick out like a sore thumb.

To avoid this situation, I would ask the following questions when thinking about technologies that are durable through downturns by yielding continuous ROI:

1. Does the product provide fundamental simplification?  Or is it a masking layer that doesn’t tackle underlying complexity?
2. How much training is involved in order to get started with the product? How much is required to achieve the highest levels of proficiency so that I can scale?
3. How many steps are needed to set-up the system versus the functionality covered. What’s my measure of efficiency and scalability?

I’d love to hear from vendors on their own set of questions. We can all learn from this discussion no matter where we sit in the stack.

Late last week, SDForum in San Jose, CA, hosted a standing-room only conference on Cloud Computing.   The speakers were a good mix of industry analysts, end-users and vendors with a very lively audience that asked some great questions. There were several key take-aways from where I sat…most notably that Cloud Computing is part of the natural evolution of computing environments and trends that began with SaaS.

Yes, Cloud can be disruptive but that won’t come from pure adoption. The disruption will come from significant advances in the underlying infrastructure, heterogeneous management systems, capacity/workload management and the emergence of open standards.

James Staten of Forrester Research forecasted that the hype bubble around Cloud Computing will dissipate toward the end of 2009/beginning of 2010. Why? Because there are few real examples of profitable solutions solving real world business problems. The bulk of the conversation at the conference was on the “cool stuff,” that Cloud might support in the future, such as gaming and mobile computing.

And that leads to the second take away - it’s all about the infrastructure.  The absence of discussion around what is the right IT environment, data center architecture, DR and HA requirements, policies and management tools to ensure that Cloud service providers can meet SLAs was noticeable.  Without the right virtualization and I/O management capabilities to simplify administration, server provisioning and application deployment to ensure availability, scalability, and disaster recovery, Cloud can not realize its potential. We’ve truly come full circle - and it looks like technology innovation is going to get its rightful spotlight again!

And that leads to my third take away…the need for standards.  Users of Amazon’s Cloud can’t move their applications to Google’s cloud and vice versa.  That’s a problem and the audience called it ‘vendor lock-in’ and you know what? They’re right.   The wider vendor community needs to come together and start to agree on some basics.

Let me know your thoughts on Cloud and the challenges we have to make it real, and then make it work.

Tags: Cloud, Cloud Computing

Over the next several posts I’ll be exploring the impact the financial crisis is having on all of us - vendors, customers, individuals and leaders. I hope we’ll get some good discussion going - and be able to openly explore how we all can manage through this. After all, challenges breed opportunities.

The financial market troubles have changed perceptions, strategies and expectations 180 degrees in a span of hours.  Many say what was once considered prudent course of business is now perceived to be reckless, as enterprises, households and individuals ‘batten down the hatches.’ Let’s face it, we’re all unsure of the best moves to make at this point.

While we can’t ignore the fact that the overall economy has caught a case of the flu, let’s separate that from what’s happening within the IT sector.  Commerce, the buying and selling of stuff, is not going to stop.  Competitive actions aren’t going on hiatus. The fact is that IT spending will continue to be driven by the strategic objectives like globalization, aligning IT with business goals, and increasingly by infrastructure initiatives to increase agility, reduce complexity and maximize ROI. And although overall IT spending is expected to slow, as we look forward in time, data center spending and the energy it consumes will account for an increasingly larger percentage of the overall IT market. The bottom line is, things don’t just stop.

IT organizations, under sudden budget pressures, will change how they acquire new technology. They have shifted from experimenting with new ways to solve emerging challenges to, now, implementing proven solutions that demonstrably address discrete business issues. And they want these proven solutions from vendors that deliver what they promise and are committed to making their customers successful. That doesn’t mean companies will opt for the ‘safe buy.’ Safe buys aren’t always the best solutions.

So while scrutiny on IT investments just increased significantly, this actually opens tremendous opportunities for management to re-evaluate their investments and solve today’s problems with the best solutions. The need for proven disaster recovery, high availability, easy-to-use management tools that support heterogeneous environments has never been higher.  Now is the best time to develop a plan and implement a strategy to achieve a reliable dynamic data center.

Because when the economy begins to roar back, you’ll be ready.

In my last couple of posts, I discussed the journey that we are on and how applications will be purchased and deployed going forward. In this post, I’ll discuss how I see systems being provisioned to support these applications.

In order to support an environment where applications are purchased as a service, with performance and availability attributes, the underlying platforms must be very flexible and have advanced capabilities such as Quality of Service and High Availability.

First, on the flexible part… it’s no longer good enough to buy hardware based on the application to be developed. This is the old model and frankly it got us where we are today - with lots of underutilized, inflexible silos of hardware. The new model mandates hardware as building blocks. This doesn’t imply that one size fits all - in fact, nothing can be further from the truth. What it does imply is that the hardware can be identified and configured automatically and remotely. It implies that hardware can be re-purposed for many uses. And finally, it implies that hardware, when identified, will be able to communicate its attributes so that the appropriate applications can be deployed to the best-fit hardware.

These hardware attributes can be as simple as # of sockets and memory size, or can be as complicated as - does the hardware support redundant power and redundant devices, does the hardware support RDMA NICs or QoS capabilities, etc.

This is obviously a big change to how applications and their host servers are deployed today. However, the underpinnings are there to support such a deployment method. 10G Ethernet is starting to gain traction in the data center and this can be used as a converged fabric (see Sadry’s post on this). Hypervisors are starting to flow QoS networking information up to the guest machines running on them.

So, the underlying platforms have the basic capabilities, but those alone can’t make the correlation between the applications and the servers. This requires some advanced policy and/or automation software, which can understand the business process, the application requirements, and match those to the best-fit underlying hardware platform.

So, there you have it. That’s how I see applications being purchased, deployed, and run on a dynamic data center environment. Sounds a bit like the cloud vision, but it is a bit more complex and mission critical than your average cloud, which tend to like homogeneous hardware and an all-virtual environment.

Thoughts? Leave a comment or send me an e-mail.

My earlier post focused on my general observations about VMworld and the hot topics in play. Needless to say from inside and around the Egenera booth, I was pretty overwhelmed by what a great reception our technology and our vision garnered.  Incidentally, this was our first year with a standalone booth at VMworld and it was great to meet the many data center managers who said “Wow, I’ve heard of you guys, and now I get to see your software and hardware in action.”

The more they saw how it worked, the more they liked it. I know I’m supposed to be modest but it’s hard when you’re coming off such positive feedback. I think people were pretty excited to check out our Capacity-on-Demand preview in action.

Our SE team started with a demonstration of exactly how easy it is to create and deploy a server onto the hardware using the PAN Manager software.  When they saw the server booting up just after a few mouse click set-up, their attention became more intense.  PAN Manager software makes that possible - and the fact that we can guarantee seamless failover, plus deliver up to 70 percent reduction in capex and opex for the data center - well, suffice to say those are strong differentiators for Egenera and our customers.

The pièce de résistance was the automatic failover.  The SE would walk them over from the PAN Manager software screen to the hardware, and then proceed to take out the blade from the chassis (we happened to be running PAN Manager on our own hardware for this demo).  A flashing blue light reappeared on another blade, and they walked back to the PAN Manager screen.  Voilà – the same server is automatically rebooting.  They were even more amazed, and it evoked further curiosity and questions – how did you do it, does it work with ESX, can you bring over the virtual machines, etc. Yes, you can do that with ESX, and PAN Manager software also runs on Dell and Fujitsu Siemens hardware so there’s choice and flexibility along with performance!

PAN Manager delivers today on the dynamic data center vision that most vendors are only talking about (anyone else noticed a lot of “journey” talk lately?), and has been for seven years in production. I’m glad to see the realization that this is possible NOW begin to catch on!

Seeing is believing.  Never more so than with changing the paradigm towards the dynamic data center.  Many promise, few deliver.  Or they deliver something even more complex, which, let’s face it, doesn’t resolve anything.  It takes a demonstration first to show how it really works.  Then we get into meaningful discussions about the details – computing capacity, training, expandability and so on.

PAN Manager software’s rapid ESX deployment and Capacity-On-Demand capabilities drew a lot of attention as users were looking for ways to accelerate their projects and optimize their hardware usage.  Users with (or planning) large environments see these as necessities not “nice to haves.”

On a final and lighter note, the attendees loved the orange Egenera bags.  It created a lot of buzz and everyone wanted one.  The whole show floor was glittering with orange. It happens to be my favorite color too

Tags: Capacity on Demand, Dynamic Data Center, Egenera, ESX, Virtualization, VMware, VMworld

It almost felt like back to school for technology last week at VMworld – a lot of excitement and movement in the industry. Not to mention, more attendees, more vendors and more vision. I had a blast!

Predictably, everyone I talked to has some deployment of virtualization in production.  Of course, they were familiar with virtual machines and their basic operations and value.

So what was interesting was that many people were strolling around the show floor looking for solutions to help them better manage their “virtual environments.” Not just their VMs. That was the dominant theme this year. I was pleased to see users thinking more broadly.

There were many case study sessions illustrating, “how I did it and what I had learned”.  Technical sessions and presentations focused on features and products that deal with increased efficiency such as thin provisioning and streaming of VMs from a single image, and single logical domain presentation of network channels to reduce set-ups.

There were the dozens of pure-play vendors, large and small, targeting basic manageability – network discovery of VMs, groupings and policies, event correlation to find the root cause of performance problems, finding “abandoned VMs” that were created eons ago but haven’t seen the light of day since.  The predominant sales pitch? Control VM sprawl – “you can only manage it if you know you got it!”  This was best illustrated by the programs that used a Google-like search engine to discover what you really have.

The strange thing is that almost all these management solutions just focus on the VM layer in isolation without considering the other servers still running physical that power them from the depths. Have we really become so solely-focused on one layer in the entire data center?

The IT world has long implemented the multi-tier application paradigm with light and focused application front-ends and mid-layer components passing information to heavy duty databases and transaction back-ends.  The latter is still best run on physical – they don’t need any extra overhead, and they don’t suffer from the drastic underutilization problems that this round of VM server consolidation is targeting.  So in my view, the “VM-only” management solution users are still only getting half the picture and less than half of the value they likely need.  VMware’s b-Hive acquisition promises to address a part of this – the end-to-end traffic analysis capability is a good start for customers.  Integrated physical and virtual management would be even better.

One application that still hasn’t crossed the chasm is DR.  There is a palpable hesitance when those two letters are uttered.  I don’t know whether it’s because of inner guilt (as in “I know I should have a more solid DR plan and technique”) or conditioned fear (as in flashbacks to performing tape restore and crossing your fingers that it will somehow reboot, or justifying the expense of another layer of idle equipment and software to CFOs).  But it goes to show that for DR adoption to become mainstream, it has to come right out of the box, with no extra knitting and weaving set-up, and no idle equipment.

Those were my first thoughts on what was hot at VMworld. I’d love to hear your own observations!

In my next post, I’ll talk a little more specifically about Egenera’s presence at VMworld – what we showed, what we learned and how the folks we spoke with reacted. Stay tuned! And as always, we welcome your feedback.

Tags: Disaster Recovery, Egenera, VMs, VMworld

It’s my turn to blog and there is a lot to discuss.   Let’s start with the notion of a truly reliable and responsive IT function.   As a CEO of a relatively large company, it’s easy to define what the ideal IT function looks like.  I want an extremely reliable systems environment.  I want an environment that can respond quickly to new business ideas or changes in the business environment.  And I want it to be cost effective.  Simple, right?  The answer unfortunately is just the opposite – today’s IT function is very often cumbersome, complicated, slow and expensive.

The next natural  question is,  why?  The primary reason is legacy architectures and our unwillingness to let go. The way today’s complex systems environments super glue together IT hardware (servers, storage, networks) requires them to be unglued and re-glued by IT staff when business needs change.  In larger organizations, when you have the added challenge of involving multiple  departments, this literally takes months - and very often, even longer.  Everyone agrees that there has to be a better way.  Not a new tool or piece of software layered on top of a currently complex infrastructure, but a completely new and innovative way of setting up and managing IT environments.

Solutions are available today, but our legacy handcuffs are holding us back.  It’s time to break away from the cuffs. We can’t afford not to.

One new approach  is to manage computing infrastructure much like what SAN or NAS did for the world of storage.  In the past, the storage legacy handcuffs were called JBOD ( just a bunch of disks).  We unlocked the JBOD handcuffs and as a result deployed a much better way to manage, share and protect data with the introduction of SAN.  Today’s legacy server handcuffs is JBOS (just a bunch of servers) with all the same issues of JBOD!

The business objective is clear - it’s implementing a “reliable and dynamic data center.”  This new world  has some enabling technologies in the area of infrastructure virtualization and automation.  And the key that breaks the handcuffs is  called “PAN” or “Processor Area Network.”  The industry is moving in this architectural direction with nearly every vendor including this in their vision – and a few are delivering it today.  But to implement it correctly, we need to unlock the legacy handcuffs that always, always hold us back. And this means re-thinking today’s hardware infrastructure architecture, how it utilizes our expensive assets including equipment, floorspace and the power it consumes, and the people resources required to manage it, with the goal of creating a dynamic data center that is reliable, responsive and cost effective.  I think most business leaders are ready, willing and able to shed the handcuffs and support a reliable, dynamic data center.  So what’s holding IT back?

Share some of your thoughts or drop me a line by e-mail.