Last week the analysts at ESG published their results from tests they ran on Akorri’s BalancePoint product.  You can download a copy of their report from here.

The guys from ESG were great. It’s clear they’ve done this before!  They showed up at the Akorri facililty in the early morning and wanted to do everything from the initial plug in of the appliance to running the most advanced analysis the product offers. Of course Murphy did have to raise his ugly head and when they plugged in the appliance it tripped a circuit breaker. Lesson learned! It’s not a good idea to plug the appliance into an outlet that already has a storage array, switch, and a few servers already plugged into it.

Ok, I know its been a long time between postings but its been a real hectic summer! June was filled with industry shows and customer meetings along with closing business for our Q2.  July hasnt been any less hectic.  The main focus for us in July has been the introduction of our latest release of BalancePoint, V1.7.   This release enhances our support for virtualization technology from VMWare and Hitachi.  It also adds support for NetApps Fiber Channel and iSCSI interfaces.

I did take a week off in July to vacation with my family at Sunset Beach, NC.  My wife might disagree though that I actually was on vacation since my cell phone was on constantly.  If you like warm water and nice sandy beaches Id highly recommend this place. Theres really nothing more to do than sit on the beach and soak up the sun, unless of course youre back at the cottage on your cell phone!  I promised her that next year I wouldnt bring it!

One of the best trade shows for Akorri this year was the Gartner IT Infrastructure Operations & Management show held back in mid June.  Our booth was over-flowing with interested attendees and our VP of Marketing, Tom Joyce, gave a talk on virtualization best practices that was very well attended.  You can see Toms presentation here.

The point of Toms presentation was to say that if you have one or more of the 2.3 million virtual servers that IDC estimates were deployed in 2006 or if you belong to one of the 76% of enterprises that plan to deploy virtualization within the next year then you need to think about how you are going to plan, deploy, and manage that infrastructure.  First and foremost youll need to look at your entire virtual infrastructure as a complete system.  Youll need cross-domain analysis, an understanding of your infrastructures performance not just component level performance, youll need to understand how virtualization makes everything shared and that sharing causes contention points and hot spots, and youll need to understand how to dynamically load balance your infrastructure in order to meet performance and availability metrics.

If you are serious about deploying a virtual infrastructure then give thought not only to which virtualization vendor you are going to deploy but how you are going to manage that infrastructure in the long run.

In early 2005 as I started Akorri I knew I’d have to somehow discover and collect information from resources within the enterprise data center. Initially my plan was to partner with vendors that already did collection and layer the Akorri value on top of that without reinventing the wheel. Great idea huh? So I began the discussions to partner with a couple of the SRM vendors for our discovery and collection needs. The plan was we would augment their agents to collect the performance centric data that we needed. Everything was moving along nicely until both of the vendors I was talking with got acquired. Shortly afterwards I found out that the acquiring companies had no plans to offer API’s for these services. Not such a great idea now! I made the decision to develop our own discovery and collection service. Although it added to the demands on my engineering team it gave us the control to develop a premier agentless discovery and collection system for performance management. That was one of the best decisions I’ve made. As development of our discovery and collection engine started I asked my engineering team to support SMI (Storage Management Initiative). In late 2005 SMI looked promising and I was a huge advocate. Having a standard interface for management of storage components makes a lot of sense. The networking guys figured that out decades ago and it’s served them well. Anyway, my Engineering guys came through and developed an SMI collector that was one of the best in the industry.So roll forward 12 months. Our BalancePoint product is now Generally Available and we are deployed in a number of enterprise data centers. Each time we engage a customer we try to use our SMI collector where ever possible. The problem is it very rarely works. We’ve found that in a lot of cases the incumbent systems vendors haven’t deployed their SMI solutions in these accounts, or if they have, the SMI technology is very slow or worse yet it doesn’t respond at all. It’s become clear to me that the vendors are paying lip service to SMI but in reality aren’t supporting it.Recently we deployed into a customer site that contained a fair number of storage arrays from a vendor that has traditionally been a strong supporter of SMI. My field guys took this as another opportunity to try SMI. I figured it was worth the effort; if we were going to have success with SMI anywhere it should be with this equipment. Wrong. It worked as well as other implementations I’ve come across, meaning it didn’t! Queries took hours to complete! After trying to get it to work for a couple of days we were put in touch with the equipment vendors field guys. His advice was to give up on the SMI interface and use their proprietary API. We made the switch and things began to work much better.  I’ve been in technology for 25 years and I’ve seen this story before. SMI is a great idea but great ideas very rarely win in the marketplace. Good enough at a reasonable cost is what usually wins. In order to succeed the equipment vendors and customers have to be convinced that SMI brings real value. Right now though it’s the good enough custom implementations for management interfaces that are winning. So my question is whether SMI is dead or just in the deep trough of disillusionment? I think it’s dead. I can’t see the magic bullet that’s going to cause the equipment vendors to change their views and truly start supporting SMI. Customers aren’t yelling at them, vendors like Akorri don’t pull enough weight with them, and the heavy implementation costs associated with SMI aren’t very attractive. SMI can join other great technologies like Beta Video Tape on the List of Great Technologies That Didn’t Make It.

Wow, I cant believe its been a month since my last posting. What a month its been. Startups are intense places and Akorri is no exception. Since our company & product launch on January 15^th Ive been on a whirlwind tour of customer and partner visits and conferences.

On March 2nd I was asked to speak at the New York CMG. For those of you not familiar with Computer Measurement Group (CMG), its a group of professionals who are interested in performance measurement and analysis of computer systems. CMG is a great resource if you are interested in computer systems performance and Id highly recommend joining.

The NY CMG chapter meeting on was focused on virtualization. Virtualization is a hot topic these days and its impact on system performance is a difficult one to understand. My talk focused on a few key points.

The first point I made was that virtualization is not new. Networks have been virtualized for years. You could also consider things like server based volume managers as a virtualization layer. And lets face it, most storage vendors have had some type of virtual LUN for years. Its server based virtualization like VMWare and Xen that are causing all the hype these days.

The second point I made was that what virtualization does for an IT organization is that it turns physical resources into services. VMWare for example turns computes into a service. If an application needs more computes then we turn a dial in the server virtualization layer and get more. If a storage volume needs more space or different protection characteristics we can turn a dial in the storage virtualization layer to achieve this goal. This is a very powerful realization. But as you can imagine, it also has a huge impact on effective system management and overall system performance.

My third point was that in order to achieve effective management of a virtualized IT infrastructure, a new management paradigm needs to be embraced. Virtualization can provide for highly dynamic systems, systems that can change relatively easy with changes in business requirements or workload. But in order to realize the benefits of this dynamic system, a high degree of collaboration between various organizations within the enterprise needs to exist. The business owner, application administrator, server administrator, network administrator and storage administrator all need to be in sync as to defining the requirements and how those requirements will impact their areas of focus.

In a future post Ill talk in more detail regarding how I believe that the available collaboration tools known as Web 2.0 can be used in achieving effective IT management in our virtualized world.

A recent Byte and Switch article described a communication gap facing a particular storage architect: CIOs measure storage utilization while storage architects measure on performance. Michael Passe of Caregroup spent a lot time trying to reconcile these two worlds. His real challenge was finding the right management software to achieve his objectives.

Describing this challenge, Michael said: I think you end up spending most of your time managing the product rather than managing the environment.” Michael is referencing the fact that his management applications have become so complex and that they require more people and more time to manage than the actual management of the services and resources themselves. I cant tell you how many times Ive heard that from customers over the past few years.

Its my belief that consumer based software products have raised the stakes for enterprise software vendors.  Customers use Exchange, Excel, and other business productivity products every day in their office environment. They are finding these products easy to understand, start, and navigate.  Whether consciously or unconsciously they now expect, even require, that the enterprise software products they purchase have the same Ease of Use (EOU) properties.

When I started Akorri this was on the forefront of my thinking.  I needed to develop products that are as easy to understand, deploy and use. Clear, simple goals that are actually very difficult to achieve.

The gauntlet has been thrown down to enterprise software companies and we better listen. Heavy frameworks that require a team of people to manage and use just doesnt cut it anymore. Customers are demanding that the products they buy put more time back into their day, not take away from it.

As I alluded to in a previous post, maintaining existing service levels, adding new resources and services, and troubleshooting performance and availability issues in the increasingly complex IT environment are difficult enough challenges on their own. Combine that with budget constraints and your average IT manager faces a quandary: How do you roll out new services, continue to maintain existing service level commitments, and stay within your budget? Put another way, which will you get called to the mat for first: an application performance issue, or cost overruns? More importantly, how does an IT manager keep from getting called out on either of these issues?

Now you may say that cost really isn’t an issue anymore, what with storage becoming so inexpensive. StorageMojo had a great post back in October disspelling this myth: yes, the cost of storage capacity has gone down dramatically, but in that same amount of time, the cost of storage processing has doubled according to Robin:

In storage, the capacity illusion reigns supreme. We measure storage utilization by looking at capacity in gigabytes, which, as Hu points out, is the cheapest part of storage. The expensive storage component is I/O. And the expensive management component is people.

Five years ago, the average disk drive cost about $4/GB while the average cost of OLTP tpmc was about $20. Today, 3.5″ disks are about $0.30/GB and OLTP tpmc is about $4. So capacity is less than 1/10th the cost of five years ago, while I/O is about 1/5th the cost. The relative cost of I/O has doubled in the last five years.

This tells me two things. First, we’re all probably underestimating the true cost of storage. But secondly, and much more importantly, it highlights the need for storage management systems to track more than just capacity and application-specific utilization. To be successful, an IT manager needs to supplement SRM reporting and management tools with other systems that can, for instance, understand and predict how the complex relationships between applications, servers, networks, and storage will affect business performance and availability. These new products must predict issues before they impact business operations and help properly budget for future capacity growth given projected application demands.

This kind of insight requires a cross-domain approach–it requires close cooperation across storage, application and infrastructure management systems. It’s another reason why I feel that the time is right for Akorri.

In 1999 I founded Pirus Networks. A technology shift was occurring back then that I felt I could bring some value to. As enterprises shifted from direct attached storage to storage area networks (SAN) I felt it was also a good time to re-examine how data services where being offered in the data center.

At that time all of the data services–like data migration, data replication, file services, and virtualization–were being done at the end points: either the server or the storage system. With SAN technology firmly in place I felt it was a good time to bring those services into the network. Offering SAN based services made a lot of sense, it was a single point of management, scalability was a lot better, and it provided an easier way of bringing data services to heterogeneous systems.

In 2002 Pirus was acquired by Sun Microsystems. It was a great acquisition and Sun was able to deliver the StoreEdge 6920 system powered by the Pirus product. At Sun I ran the Storage Platform Group. In this role I heard first hand the issues customers were facing. The biggest issue I heard was one of complexity. Sure, it’s great having all of these services available–after all, having flexible services with virtualized physical resources is what will make the On-Demand data center a reality. But, as most IT administrators admitted, managing those services and resources can quickly become a nightmare.

Questions such as, How should I virtualize my server or storage to optimize my applications performance? or When and where should I migrate my data? or Is my infrastructure configured properly to support my application portfolio? were all questions that IT folks felt need answering.

After leaving Sun in 2004 I continued to think about this problem. It occurred to me that the only way to truly solve this complexity issue was to fully understand the cross-domain (i.e., server, network, and storage) technologies that 1) make up the data center, 2) characterize the applications requirements for the infrastructure, and 3) use advance analytical techniques to pull it all together.

I realized that if I could characterize, or fingerprint an application and use that fingerprint information against a mathematical representation of the infrastructure then I could bring the knowledge needed to effectively control the data services and virtual resources. My challenge was to bring together infrastructure and application engineers along with mathematicians to work on this complex problem. I felt it was a perfect problem for a startup to attack. Being able to get a small, focused, diverse set of people together in order to solve a complex issue is what startups are best at.

So with that understanding I founded Akorri. With many man years of development already behind us, we’ve produced our first product, BalancePoint. BalancePoint will give IT administrators the visibility and knowledge they need in order to troubleshoot, optimize and plan around their complex environments.

I hope this gives you a good understanding of the thought process behind our company and product. In my next few posts, I’ll explore many of the different balance points that managers and administrators must find as they walk the modern IT tightrope.

Welcome to Akorriblog. My name is Rich Corley and I am the founder and CEO of Akorri. I’ve been described by many people as a serial entrepreneur. Like most entrepreneurs, I have a passion for creating and building. That passion for creating started as a teenager building ham radio’s from scratch. Nothing is more satisfying then seeing something you’ve designed and built from the ground up work.

I’ve found that building successful companies that solve real customer problems delivers similar satisfaction. The serial part of “serial entrepreneur” comes from the fact I’ve done this six times! I guess that qualifies this as an addiction for me.

Another passion for me these days is the evolution of the enterprise data center. It’s amazing to see the transformation in twenty years from mainframes to distributed systems and now to the emerging on-demand, service-orientated data center. The focus of this blog will be to explore the directions I believe the enterprise data center is heading, and to share with you some of the thinking behind Akorri’s business–which was built to help companies make it through this transformation.

I’ll look at some of the current challenges that exist in today’s data center and how Akorri can solve those problems, while keeping an eye towards the technology and business initiatives that will transform the data center over the coming years.

I’ve brought together a seasoned team of system and software engineers, world-class mathematicians and business people in order to create solutions for the issues that data centers face today and become leaders in helping in the evolution of the data center. I’m looking forward to sharing some of the insights that this great team has as we tackle some very challenging issues for our customers.

I can’t promise to be brief in every post, but I will be focused. If you’re interested in the changing world of the datacenter, or your ears perk up when you hear words like storage, virtualization, application performance, SLAs, troubleshooting and capacity planning, you’ve come to the right place. I hope you’ll come back. Thanks!