This is why I tell prospective partners that it is just as important for us to get to a “no” answer quickly as it is to get a deal in place. I would rather stop an engagement early before any real damage is done than to have to terminate an agreement with disgruntled people on both sides. We are usually very good at coming up with all the reasons why we should have a partner relationship:  unlimited revenue, big channel, great technology, etc. But we tend to forget all the reasons why we shouldn’t engage. This is why it’s so important to be earnest and forthright. It’s important to understand your own limitations as well as the prospective partners before moving ahead. For example, is there executive sponsorship in both companies? Are there proper resources such as product management, marketing, field, developers to do integrations assigned? Is there a go-to-market plan? Is there a cultural fit? Can a small company deal with the larger company? If those things aren’t in place, then how will revenue happen? I’ve seen relationships fail with everyone in place except a field engagement model and it turned out the sales people could not get along culturally – actually it turned out to be a compensation issue.

This is why I consider integrity the most important trait in business. Not only is it the right thing morally, but it’s the right thing for business. With integrity you can honestly assess with the prospective partner whether the relationship will really work. With integrity, comes predictability that your organization will deliver what they claim. And so, with integrity comes the opportunity get to a no answer more quickly than beating around bushes or glossing over limitations in your organization and thus wasting time and resources of both companies that could be spent pursuing real opportunities. At the end of the day getting to “no”, in this small world, may mean getting to a bigger “yes” with the same company or people from that company at a later time because they respect you and know that you won’t waste their time.

It has been awhile since I blogged and a lot has happened since. I’d like to say thanks for all my father, Oluf, did for me in giving me an example of a person with integrity in every aspect of his life. And being a great human being all around. He passed away May 12^th, 2008. RIP

community (plural communities)

1. Group of people sharing a common understanding who reveal themselves by using the same language, manners, tradition and law. (see civilization).
2. Commune or residential/religious collective.
3. The condition of having certain attitudes and interests in common.
4. (Ecology) A group of interdependent organisms inhabiting the same region and interacting with each other.

I can’t help but think that we’re in a transition phase with the whole Web 2.0 concept (it’s really concepts). And of course, we’re always in transition phases regardless of whether your talking about technology, society, or just life. So, what I’m saying is we may be reaching an inflection point in the evolution of social networking. We have many very interesting social networking enablers with blogs, wikis, podcasts, rss, etc. These are effectively evolutionary technologies from their roots in Usenet and ftp which I was using back in the ‘80s. All these technologies provide a shade of the definition of community but they do so from mostly an information exchange perspective. And now we have consolidators such as Jive that join many of the technologies together in one location to give an even greater sense of community by giving access to much of the information in one locale such as BMCDN.

This is why I think we’re reaching an inflection point:   basically, even with a consolidator, we’re still doing store and forward. We have elements of chat/interactivity, but we’re still operating at a fairly superficial level. Based on the definition above, we share language and we have agreed upon manners, we also have media for sharing common interests. But look at the deeper aspects: 1) People who reveal themselves by using… tradition; residential/religious collective; interdependent…interacting.

We’ve taken a very data centric approach to the online community concept, so we’re pretty efficient with data storage, search and retrieval, but we haven’t quite mastered the interaction method. A community should be a place you spend some significant time: you exchange ideas, you give, you take, you argue, you learn, you grow, you help others. And you have varying relationships: some cursory, some technical, some friendly, some public, some private. And these characteristics change dynamically.

All a twitter… I had been skeptical about twitter until I read a Wired article a couple of months ago titled, “Clive Thompson on How Twitter Creates a Social Sixth Sense”. The article is about how the aggregate effect of getting “pings” of everyday life from people you are interested in creates a sense of awareness of those people that goes beyond the superficiality of the information they’ve provided. Twitter and other constant-contact media create social proprioception. He says, “They give a group of people a sense of itself, making possible weird, fascinating feats of coordination.” In other words, it gives people a sense of community. He goes on to say, “This awareness is crucial when colleagues are spread around the office, the country, or the world. Twitter substitutes for the glances and conversations we had before we became a nation of satellite employees… So why has Twitter been so misunderstood? Because it's experiential. Scrolling through random Twitter messages can't explain the appeal… but the real appeal of Twitter is…[that] it's practically collectivist — you're creating a shared understanding larger than yourself.”

Wither now? Look at the evolution of Second Life from an online gaming type of community to a medium for online classes, marketing, and (drum roll) collaboration! Combine the shared consciousness of Twitter with the 3-d type of community that Second Life provides and you start seeing the revolution of the virtual community. Now the interaction method becomes oriented to the actors (us) and we are able to create personae for ourselves that reflect the particular social interaction that we are engaging. Second Life enables you to create avatars for yourself (in fact you can buy them online if you don’t want to create one from scratch). What we still need to work out are the filter mechanisms for the interactions – I should be able to create outbound filters that indicate things like mood, desire to interact, and level of interaction that influence other’s ability to interact with me. It’s kind of like having spheres of influence that are set in two ways: 1) statically to reflect my core type, and 2) based on circumstance and level of relationship. However, once we start engaging in a revolutionary type of media such as SL, then the evolutionary processes will kick in to provide higher and higher levels of service and nuances of relationships. The type of complexity you need to have a real sense of community.

And then we need these identities to follow us whether we are on a laptop, phone, pda, or any other interactive device. Now the interaction method becomes human-centric so rather than exchanging data, we communicate in memes – ideas that evolve and become an artifact of the culture – a new, dynamic, and growing online culture – Web 3.0?

If you want a long term projection of where this can go, read “The Golden Age” trilogy by John C. Wright.

P.S. I referred to Weather Report last blog. Joe Zawinul, the founder along with Wayne Shorter, died last week after battling cancer. RIP.

My favorite album of theirs is "Heavy Weather", which included one of the best bassists of all time, Jaco Pastorius (RIP). A real work of genius on that album is "The Juggler", which has my very favorite song ending  - a single forlorn bass note giving an ironic twist to a complex song. Jaco later connected with Pat Metheny and you can still hear Jaco's influence in Metheny's later albums with Mark Egan on bass.

It's fascinating to trace these influences and their branches into other genres of music. At each juncture, a musician influenced another musician by participating in the creation of new music, but their influence continued well after they moved on.

Before open source, the ISV world was Borg-like, “resistance is futile.  We had no choice but to absorb, regurgitate and attempt influence through traditional marketing, creating countless standards committees, and raiding each other's staff.  Software is now growing up.  Adding open source to the ecosystem creates the opportunity for collaboration between developers regardless of corporate affiliation. The resulting innovations can then be incorporated and used in powerful, sometimes unanticipated ways.  Each participant in the process is affected.

I believe it is mandatory that anyone who calls themselves a "professional" developer must be able to show examples of open source contributions - and the companies they work for should encourage this participation. These contributions become the incubator of innovations that will later find their ways into commercial applications that will help change the world.

I’m proud to say that BMC’s permissive licensing is a potent enabler of rapid development and innovation in the systems management space. Permissive licensing incents developers to build on the baseline code provided and build commercial innovation. And I believe that as such developments become visible to the community there will be natural pressure to encourage contributions back to the community. A good name is still important in the industry. In addition to this, market demand has created the opportunity for enthusiastic partnerships in the new ecosystem to help fill existing and future software needs. This can be in the area of platforms, specialized applications, and variations of current applications. Just look at all the custom applications developers have done on the Remedy Action Request System. It's important to have the infrastructure for the market to express their needs and to have those needs met with either commercial or open source software - and they are not mutually exclusive.

We have all this with developer.bmc.com. Take a look.  Participate.  Give feedback. Check out the new podcast with whurley and me talking more about BMC Developer Network.

We're so glad you could attend

Come inside! Come inside!

~From Karn Evil 9: First Impression on Emerson, Lake, & Palmer's Brain Salad Surgery Album

ELP is one of the original progressive rock bands with the original likely being King Crimson (Greg Lake's original band). Prog music is fun to trace because you can see influences inside and outside the genre even into pop music. The music industry is an interesting ecosystem that is always riding an edge between IP protection and creative license. When does a technique or a riff or lyric move from protected property to open source? And as most would agree, the music industry does not deal well with open source. :-) However, there are many examples of modern music openly using classical riffs from greats like Beethoven and Bach - so it does exist. And so it goes with the software industry - there are fits and starts with ISV's incorporating, developing, and contributing to open source but it is happening and becoming a larger and larger factor in ISV strategies. And thus it goes with BMC - we actually have, since Y2K, incorporated, developed, and made open source contributions - but generally not in a strategic context. Now we are.

In my last blog entry, I mentioned; "Such a strategy requires an infrastructure that narrows the distance between the platform provider and the developer community". This was an allusion to our BMC Developer Network which whurley announced at OSCON this week.

We've actually been semi-live with BMCDN for a couple of months, migrating the Remedy groups and also creating new forums aligned with our BSM strategy. We are also in the process of migrating the DevCon (a.k.a.: PATROL Developer Connection) to BMCDN. But, as you can tell, our release Open Source adapters at BMC is significant and, I believe, helps complete our ecosystem approach.

BMCDN is by no means perfect nor complete.  We had a choice to attempt sterile perfection or to get the word out and truly commit to having the developer community drive the structure and content of the BMCDN. We chose the latter. So we have an initial structure and we have some content, but I'm hopeful that a year from now we'll have something completely different that has been tailored and built in conjunction with our development community.

I would like to acknowledge the hard work over the past year by people on my team including Ken Beck, David Fiel, Joe Vodvarka, Scott Powell, and Luis Laborda in getting this BMCDN ready. This not only included development of the infrastructure, but also included getting the open source adapters built, licensing schemes debated and agreed, and coordination with R&D and Marketing.

I'm excited about what we have and more excited about where our development community will take it. Come and join the show -- the BMC Developer Network developer.bmc.com!

Such a strategy requires an infrastructure that narrows the distance between the platform provider and the developer community to the point where there is near seamless interaction with all elements of the ecosystem - joining customers, developers, and the solution provider into one dynamic bazaar. I think this is where ISV's can learn a lot from the open source community...

BSM solutions, while being comprised of major components built by BMC, still have partner componentry as part of the whole solution. For example, BMC has the leading Knowledge Management for Service Desk in the industry. Some customers may want to purchase off-the-shelf knowledge to get their knowledge-base started. We work with partners to provide the knowledge content that populates our Knowledge Management solution. Same goes for Network Configuration, Mobility, Notification, and other areas that complement our core solutions.

We are being very aggressive about building out the ecosystem across the solution set and have some pretty exciting and new programs coming out around developer enablement and communications and open source. I’ll expound more on these areas later…

The bottom line is that BMC has a complete BSM set of solutions and these solutions have a healthy ecosystem around them. BSM presents, if you will, a virtual platform for enabling a whole solution that includes best-in-class for every aspect of the infrastructure from hardware to network to application to business process. Thus I see my role as having changed from focusing on virtualization technology to helping to build the virtual business – the ecosystem around BMC’s solutions.

As VMware has become more and more prevalent as a platform, the platform itself is becoming less interesting - it's there and it works - while the applications (usage of the platform) are becoming more interesting. VMware is at an interesting inflection point where they have been the focus for enablement of their platform and have provided interesting applications to migrate, consolidate, and manage the environment - and the focus (and rightly so) has been on the ecosystem around enabling the platform. And while the platform is currently dominant, it's key to have the platform specific applications that fully leverage it's capabilities. Easy examples are VMware appliances that provide specific capabilities - I mentioned the Media server in a prior blog - other examples, include demo platforms, system management appliances, and even application servers where you want instant portability. I'm curious how many third parties are making significant revenue providing such specific applications? One I know of is Surgient which provides "virtual lab management" software on "standard" virtual infrastructure (ie, VMware). From my experience, most of the current uses of VMware are around proprietary customer applications and migrating these applications from physical to the virtual environment. And that's not a bad use, just doesn't create a long term and dynamic ecosystem. For BMC, we already provide specific management functionality for VMware, and we're seeing use/revenue for these products (monitoring, performance management, and capacity planning), but almost exclusively in the context of the migrated legacy apps. I'm not saying the VMware specific applications are not there, I just have not heard of a lot of revenue being generated in that domain yet. Am I wrong?

Regardless, as the platform matures, there will be extensive opportunities - capacity planning will become more critical for production VMware workloads, power and coolling (which Dave Wagner has written extensively about) will become more critical, and we'll see a chaos of innovation we haven't anticipated.

Another aspect of containers applies to the end-user. An early example is Citrix providing, effectively, multi-user access to Microsoft Windows environments. A new example is with Softricity (just acquired by Microsoft) providing application streaming. In both cases it’s the idea of providing end-user functionality (either OS or App) from a central location (a container) that can undergo all the container actions described above. A nice aspect to these types of containers is having near absolute control of the environment and application configuration provided to an end-user. So along with that comes security to the enterprise in knowing that the environment and applications being used by their end-users exactly meets specifications. Another use case for user environment containers is providing a guest environment – give a guest/visitor access to limited applications such as web access within a “contained” environment. Then, when the guest is finished, the environment can be disposed without impact to the enterprise.

In both container cases, the environment is really looked at as a file. Both VMware and Microsoft have their proprietary virtual file formats, but, in the end, it’s just a file. This means the containers themselves, while undergoing the above operations, should also be considered candidates for Change and Configuration Management processes. Such an approach might greatly enhance the way that applications and user environments are managed thru every ITIL process. I do think there’s room for standardization of the virtual file formats – which would make containers portable across virtual environments. I doubt this will happen anytime soon, but it may be an inevitable development if virtual environments follow the same path as processors and operating systems.

I’ll end on a personal use case. My son helped setup a media center where we stream photos, video, and music from a central server in our house. We therefore needed a media server that would feed the streamed content. Well, we decided to load Windows Media Center into a virtual machine under VMPlayer. We did this because WMC needs a machine to run on and by being on my son’s laptop so we could test and tweak in the same location as the media center. Once we got it working, we copied the .VMX file to the primary server environment, and, it ran without a hitch! This way we could verify the functionality without creating any risk to the main server environment. An interesting side note, we subsequently had the hard drive crash on the primary server and recovered the media server by copying it from the backup hard drive.

One of the greatest stealth tools in the business is instant messaging (IM). IM is integral to geographically dispersed organizations – it's handy in getting a quick answer to a question while you're on the phone with someone else, coordinating and facilitating online meetings, and keeping in contact with family and co-workers while on the road. Amazingly, the IT industry has not taken a productive stance relative to IM – either IM's are ignored, forbidden, or an IM is chosen that has no connectivity to outside the firewall, which sounds secure, but generally results in noone in the company using it – which is secure, but not productive. Our world has blended the concept of work and personal life where a person can be simulaneously engaged in both from anywhere in the world – home, restaurants, hotels, and (heaven forbid!) the beach.

AIM (AOL Instant Messenger) rolled out two AIM bots called "ShoppingBuddy" and "Moviephone" placed on your AIM buddy list under "AIM Bots". To the consternation of many in the industry, AOL placed these bots without asking the AIM users (AOL forces friends on AIM customers). While I don't like the idea of any software being placed on my system against my choice, the concept of AIM bots is pretty interesting to me from a support perspective. The bot concept is about thinking, for example, I want to go to a movie, then telling someone (a bot) that you want to find a movie and getting assistance in a more human way in finding the movie, time and location. I'm pretty sure that the AIM bots are based on "Smarter Child" which provides a framework for generating conversational rules for finding information. It's a way to interact with a service in a human-like way. In fact, a properly constructed bot would likely pass the Turing test.

Think from a software support perspective:

1. You have a problem with ABC software
2. You click on the IM bot for ABC software (that you opted into when you installed the software)
3. The Bot validates your registration and asks you to describe the problem
4. Simultaneously, the Bot has generated a trouble ticket, alerting customer support (based on your support level), and uploaded configuration data related to ABC software on your system
5. As you go through the questioning process, the Bot realizes that it is not able to solve your problem immediately, so during the questioning process it raises the alert level
6. A human from ABC software takes over the IM session and drills down on the issue and finds a solution
7. The problem fixed, you're happy and running the ABC software, never knowing whether you were messaging with a human or machine.

All that matters is you got help instantly and effectively.

To get to this scenario, we need standards and security. However, much of the technology is in place including IM, support knowledge-bases, and a framework for building human-like conversations. Just seems that support is such a huge part of the End-user Quality of Experience that we need to get better and fast! And we have to admit that IM is integral to doing business in the On Demand world.

What is the most likely factor when a change (of any kind) has a detrimental effect on the enterprise? For those of you who liked, Cool Hand Luke, “What we have here is a failure to communicate.” The enigmatic nature of change is that the very communication of change tends to bring up forces that oppose it. On a side note, I think I tend to embrace change easier than most, since by the time I graduated from high school I had attended 12 schools in 6 different states and countries! There is a kind of universal nature to the change process:

Funny thing is that the change process tends to be done effectively in certain silos of the enterprise, but not really at the enterprise level. For example, most companies that run mainframes have very stringent change management processes for that platform. In fact, one of our products, CHANGE MANAGER for DB2, actually implements the above process for database schema changes and migrating them from dev to test to prod. However, the failing (not of the product) is that the process tends to be focused on technology rather than on People, Process, and Technology. Any change to one of the three WILL have an impact on the other two. Of course, having been an architect in my past, I always search for universal truths such as a universal CHANGE platform that incorporates the three. The first tendency is to throw technology at the issue because technology/tools are tangible expressions of the desired outcome – but, see unintended consequences. The very nature of introducing technology is Change itself.

This gets back to my surprise at the lack of maturity in change management processes – because, if the processes were mature, any introduction of technology to the enterprise would be preceded by a change process that factors impact to people, processes, and other technologies. And the Discovery process wouldn’t be such a mystery, because people and process impacted by change would be readily available in a mature change management process. I had a pretty unique experience a couple of years ago when it appeared that a BMC product for which I was responsible was causing an outage at a pharmaceutical company. This company has robust change management process (not necessarily technology). I got introduced to their change management review meeting which was a worldwide video conference that included business owners of the impacted businesses, IT management, and their vendors at once. The meeting had an agenda for considering changes and issues that impacted the business – and we were on the agenda. It was pretty intimidating at first, but I realized that it wasn’t a witch hunt, but truly a forum for identifying impact and workarounds from all three perspectives. Luckily, after two of those meetings, we had a workable solution in place… I guess not too surprising that a pharmaceutical would have strong change management processes – in fact I’d be worried if they didn’t.

So CMDB gives you a mechanism for storing the relationships between people, process, and technology relative to the business. The next step is to use this information combined with intelligence to do impact analysis. And, of course, we have automation to execute the changes.

Now to RSS. Using RSS from a support perspective is effectively using it as a change notification tool – a device for communicating change. I believe most unintended consequences are caused by lack of communication. RSS is a tool that can be used to communicate changes to interested/impacted parties both inside and outside the firewall – if we had a mature implementation of private RSS. Of course, the first thing to address is process, but enhancing change communication is a near first goal also. And RSS does not address how the impacted parties communicate how the change will impact them – there any many methods for addressing this already built however.

Thanks to the folks who commented on my last entry regarding private RSS, indeed there is work going on with private RSS (see link dated July 2003), but looks to be in early infancy. And I still think the use cases I outlined are fairly unique so far. Another aspect to RSS that I intend to expound further in my next blog is Change Management. RSS is, in a sense, a way of subscribing to notification of changes. I’m pretty surprised at the lack of what I’d call coordinated change management across the enterprise. And then look to further writing on the RSS software support use case. I think support is something that’s still stuck in the ‘90s even with self-service and knowledge databases – I think we’re (the global ISV community) missing the boat on managing or enhancing the customer experience. At the end of the day, positive customer experiences are what keep customers.

It was interesting to me that this trip, my phone and blackberry just worked – whether in England or Holland. I didn’t have to select networks or anything – they just worked on demand. I’ve had an interesting evolution with the hotel I stay in Amsterdam. My first trip staying there 4 years ago, I had a dial-up connection that ended up costing me over $100 in one night! That evolved to high-speed wire fee-based ethernet, then this trip, it was free wireless. Access is turning into a commodity – I just expect it to be there and to work. You can see this evolution reflected in our connection to eachother. While I was at Gatwick waiting for my flight to Amsterdam, I was exchanging e-mails with Ariel Gordon, one of our CTOs, about virtualization. Ariel is based in Houston, but our e-mail exchange was at 3:30 am Central time. Ended up that he was in Israel, but it wouldn’t have been out of the question if he had been in Houston. I later called a guy about fixing my sailboat and, as far as he was concerned I was in Austin. This whole high-speed communication thing has changed the concept of locale. Even though I’m in Europe, I log in to Austin/Houston machines, my home page displays local weather and news, my IM and Skype show people I want to connect with and their availability. So I have a “home” connection. On the other hand, I got a call from my credit card company trying to sell me more services. The call was from a guy in India. While his “locale” is India, he’s adjusted his timeschedule to match the locale of the geography he serves. When I worked with GE, I did the same thing. HQ was in New York, but I was based in Seattle, so I came in at 7:00 am or earlier, so I could maximize overlap with the folks in NY. Maybe the earth is flat?

What does this have to do with utility computing? Well, it appears to me that now I require all my services to be available all the time, anywhere. Multiply me by millions. This means that BMC Austin/Houston can’t take an internal system down during “off hours” to run maintenance. It means that utilization rates on computing systems of global oriented companies will go up because they will be used 24 hours a day (internally and externally). In fact, I wonder why, for example, stock trading is limited to “working hours” of the locale of the trading exchange. It seems like, at some point, we’ll have a major stock exchange go 24 hours a day at some point. Think of the implications for traders, they’ll need to have people available 24 hours to make sure there’s someone to sell at the point a major event (or disaster) occurs. So now resource scheduling, maintenance, and availability needs to be managed from a completely different perspective. I don’t think anyone has fully thought this thru…

If you’ve read this far, here’s the real interesting fog induced rumination. In fact, if no one is has mentioned this before I claim it for BMC… I pay all my bills online, but to-date, most of my bills are still provided only via paper. I can access most accounts on the internet and get my billing info, but it’s a pain to have to go to several different websites, log-in, and retrieve my billing info. Private RSS. Of course secure, but also private. Why can’t I subscribe via RSS to my billing statements, then use Bloglines, for example, to create one summary page that displays my bills as they change/update on each host/private page? This would simplify my billing enormously. Another use case, more in the BMC line, is support. Why not have customers subscribe via RSS to product support? Then as products notices come out, they have one page to look at. In fact, the page will automagically update. This can also be used to limit spam effects. Then any support e-mail would not have to reference a web page, the customer would just be reminded to look at their self made RSS update page. I think this would be a real interesting extension for RSS – but it has to be private and secure. Maybe it’s already done, but…

Lot’s of stuff.

Of course, outsourcers are already in this aggressive adoption phase...

Other knicknacks include getting Microsoft OneNote which is great for taking online notes and putting bullets on Outlook tasks, classifying the notes, etc. I also got a Linksys (Cisco) wireless travel router that supports G networks and also has four hardwired ethernet ports. Real handy in hotels or conference rooms with only one ethernet port for 10 people…

Now for the weird convergence... Benny gives me a call the same day I've read the article! Of course he's selling me on joining the Enterprise Grid Alliance and I'm pretty pliable since it seems that what they're doing is consistent with where BMC is heading. And I'm pretty impressed with Oracle's direction on grid. There are few players like Oracle and BMC who are relatively agnostic to the hardware platforms on which grids are being implemented and are oriented towards the application/service.

Virtualization capabilities are being built into the operating system. Red Hat and Novell are incorporating XEN into the kernel and Microsoft has indicated that they will incorporate MS Virtual Server into Vista (Longhorn) in the next 18 months. However, Linux presents a pretty compelling trend towards virtualization with it's cost benefits directly and with the fact that most people are rolling it out on commodity, Intel-based servers. Oracle is betting pretty big on Linux and Virtualization - see Oracle's Charles Phillips: On demand is the future - and using the combination to result in "Unbreakable Linux".

It just seems to me that Linux, Virtualization, and OnDemand/Utility Computing are inextricably linked. Linux provides pretty compelling licensing terms that encourages (or at least doesn't discourage) the use of hundres or thousands of commodity servers. Virtualization offers the ability to optimize the utilization of the underlying physical hardware without sacrificing availability. OnDemand ties them together to give the customer optimized use, flexibility, and availability. OnDemand also enables (or should enable) the use of the right resource at the right time. Sometimes it might make sense to provision a RISC-based box (eg, Sun or other) to handle an i/o intensive application - Intel boxes are great when you need CPU power, but that's not the only problem that needs to be solved. And the only way that any of this makes sense is to allocate the resources based on business requirements.

Power consumption and heat are major issues that IT needs to address, but, at the same time, according to industry analysts, many enterprises are running at 5-15 percent average Wintel server utilization rates (see Virtual's New Reality in Information Week). What this means is that 85-95 percent of server power consumption is wasted energy as the server sits idle. Industry analysts calculate additional powers requirements for cooling and lighting at one and a half times the direct consumption of each server. So, although IBM is providing an interesting solution, maybe the immediate problem is not power consumption and cooling, but more effective utilization of the server assets within IT?

A BMC customer was purchasing over 100 servers a month for various business applications. The CIO realized that there was little or no coordination of these purchasing activities - rather the purchasing decisions were driven via the specific "silos" within the business. The CIO took radical action and stopped all server purchases for a 12 month period. Even so, all the business requirements were met by instigating better Asset Management processes along with implementing Capacity Planning practices. The Capacity Planning practices, using PATROL Perform and Perceive,  enabled them to get a better picture on how and when servers were being utilized and to identify opportunities for sharing and/or reallocating resources. This resulted in a savings of about $5M due to cost avoidance during the timeframe. The Asset Management processes enabled them to better track where and by whom the servers were being used. Practices such as these can achieve pretty dramatic results in improving the use and allocation of assets. We call this process Asset Optimization.

Of course, one of the main reasons for this gross overprovisioning is that customers are concerned with assuring the availability and performance of their applications. The standard insurance policy is to purchase dedicated servers for addressing each of those purposes. This is normally done within a "silo" of the business. If customers could better pool their resources and allocate them just-in-time, then they would be able to improve utilization rates even further than can be achieved with improved processes. Policy-based orchestration enables customers to do exactly this and enables customers to dramatically improve utilization rates. BMC has just released BMC Virtualizer for High Availability and BMC Virtualizer for Capacity On Demand to address this issue. To quote another BMC customer, "1500 of my 2000 servers are high-priced space heaters."

Bottom line is that implementing process and practices that enable sharing of resources across IT and using tools to automate the resource allocation process can have dramatic results in reducing costs from servers, floor space, and power consumption. I think customers are better off with the more process oriented approach as opposed to the "Band-aid" approach.

VMS used this concept extensively in its session management - giving each user unique access to common services - storage, memory, processor - with virtual walls between each session/user. This became important to me in terms of being able to complete my assignments because there were not enough terminals (remember those green VT100's?) to satisfy peak demand. The student had to sign up for 1/2 hour increments of time at a given terminal. Lucky for me, I found an unused area in the Health Sciences Center at University of Washington. It seemed that medical students didn't need to sit at a terminal for 4-8 time periods to get work done. I found that I could fire up multiple sessions from one terminal and also was able to "spawn" DCL (DEC Control Language) commands to maximize use of processor time. This meant I could be creating a program in one session while running compile, link, tests in other sessions - thus maximizing use of all the computing resources. Until, of course, admins found out and gave us a 2 session limit...

Of course then I found that there tended to be fewer people in the computer lab at 2 am than "normal" hours. This worked until I got married and one night/morning, I'm hearing this knocking on the lab window above (it was in the basement) and my wife is telling me she would prefer I be home at before 2 am... This led to me getting my first "PC" a Kaypro and doing my development offline using TurboPascal (remember Borland) and MixC and attempting to assure compatibility using ANSI standards. Even with 300 baud upload speeds, this process ended up more efficient than terminal hunting...

Another funny aspect to this is at my first professional employer, my desk was in a bullpen of desks (no cubes), but the terminals were located in a separate bullpen. GE Flight Controls finally realized that developers tend to get more work done if the have constant access to the primary tool of their trade. I actually brought my wife in to the office to show off my brand new VT220 terminal with amber screen. For some reason I was reallly impressed with amber. Used same session techniques, got limited by admin, negotiated them to a 4 session limit. We hit the wall one day with computing power and brought the whole team of 25 together to discuss the most serious event of our short 2 years as a team - we needed to upgrade the VAX from 4 megabytes to 16 megabytes of RAM!

Interesting to me at the time and even 'till now was I never encountered an overlap with another user other than performance issues. The partitioning of VMS was so solid that we never tripped on eachother. Later, I also encountered VAXes running manufacturing plant control where it used the partition concept to control and collect data (in manufacturing, called SCADA - supervisory control and data acquisition) from sometimes hundreds of devices on the plant floor. And at the same time, the VAX could upload production data to "host", usually mainframe, computers at the business headquarters.

So, flashforward to today. We're looking at the virtual machine managers (VMM) like VMWare and XEN as if they are new technologies. VMS was providing the virtual machine (partitioning) concept back in 1979 and, of course mainframes before. And VMS was providing the concept of specialized partitions such as the SCADA example. Additionally, when they delivered the Alpha processor based systems, OpenVMS was positioned as a VMM that could host multiple Microsoft Windows NT sessions.

It's interesting to me that it's almost like we (the industry) forgot this and have not learned architecturally from the past. VMS was a highly efficient partitioning system. I think the lesson there is that in order to have a truly efficient VMM you need pretty tight coupling with the OS layer yet want enough flexibility to decouple the Application from the intricacies of the VMM - should just appear as an OS. Yet VMS was a building block and it had a great run. For me personally it helped inspire my career in computer science, led to my first foray into PC based computing, enabled me to develop sophisticated software for controlling amazing aircraft, I saw it running huge manufacturing businesses, and it's a model and inspiration for virtualization even today. It's amazing how you carry your past with you...

Since I'm still hobbled by the broken foot, it was a stay at home vacation.

Reading:  Canticle for Liebowitz, Ezekiel Option, Primal Leadership. First two are pre-apocalyptic/post-apocalyptic. If you like that type, I highly recommend "Forge of God" by Greg Bear and "Moonseed" by Stephen Baxter

Viewing:  Star Trek Original series. Got the first season on DVD. It's funny to see color coded 3.5" storage devices, plasma tv's, and portable communicators (that even that far into the future still have unreliable connections)... And Captain Kirk, the first blogger. "Captain's Log, Star Date..." :-)  Also watching "Into the West" a great mini-series of US expansion to the West. Gut-wrenching pieces about Native American history.

Still really frustrated that mobile telephones and home ISP connections are so unreliable. Should be like picking up the landline phone...