HP Universe

Thursday - June 19, 2008

06:30 a.m. -  I am up and at 'em. I was supposed to sleep in, but apparently my internal clock just cannot make the transition to Pacific Standard Time. The hunt for food begins...

07:30 a.m. - Tim and I make our way to the HP breakfast for the conference. It's a much more calm and somber mood. Many people do not look like they are in as good of a mood as they were last night. Hmmm. Hey folks, lighten up. Smile every once in awhile.

Everything was really calm until one of the wait staff decided to grab this mini xylophone looking thing and wandering around tables playing these ring tones. In the distance it sounded like the hum you hear in the casinos of the slot machines. Up close it became a series of tones that triggers the side of me that wants to grab a blunt object and knock the crap out of someone. Their point was to let everyone eating know it was time to leave and go to their first session. But what if you did not want to leave or did not have a first session? That was one of the few times since I have been here that I became annoyed at the conference. I am a freaking adult (well, most of the time) and if I want to have an extra cup of coffee and be irresponsibly late to my first session, so be it. We don't need to be herded like cattle. That's for Southwest airlines. 

08:30 a.m. - Attended the session, "HP Performance Center: Successful strategies for deploying performance centers of Excellence" by Greg Venable at Pfizer (also ex-Mercury FYI). OK, now we're in my territory.

The initial set of slides was given by Anant Hariharan (HP) who worked on the Pfizer project for implementing PC. He did reference the HP Maturity Model which I had hoped to see from BearingPoint, so I was pleased with that. However, he mentioned various types of CoE's, including "virtual CoE's" and "Line of Business CoE". This was extremely confusing. Either you have a CENTRAL Center of Excellence" or you have DISTRIBUTED excellence. You can't have it both ways. But Anant did not go into the details of those other CoE's so I may have simply missed what was missed here.

Greg talked about his experience at Pfizer, taking their existing LoadRunner labs and moving to not just Performance Center, but a global service offering. The one thing he wanted to focused on was that you will be running Performance Center like a "dot com" servicing its customers. I agree with Greg on this, because you have to create your own internal consulting service internally the same a company (like Loadtester) would do externally. The model simply works.

Greg first focused on implementation of the PC product and the lessons they learned getting it up and running, which was helpful. Best tip privided: Don't try to put your data on Network Area Storage. Second most important: Only use VMWARE for Performance Center when it is being tested in a test environment (at Pfizer they even had a DEV environment for PC too). Trying to use it for PC components for real does not work. Although HP says some of these components are VMWARE certified, real life usage is a different story. I am not saying it cannot be done, but it would have to be carefully planned. 

He also talked about process changes, but this information was pretty light. They had to implement a lifecycle approach by looking at information from Business Availability Center when problems were encountered in production, then moving it to Performance Center for trying to recreate the problem in a test environment. If a functional defect was found under load, it was then documented and handled in Quality Center. Pfizer has 4000 projects and Performance Center was the best way to try and tackle as many of these applications as problems and to run their department as a service.

Eric Shumacher asked how they calculated ROI. Pfizer used hardware savings (40%), number of people (reduced 35%), and better testing results. The applications teams were getting better results by using the centralized team. Eric also asked about a charge back model. Pfizer did this by creating charge codes so that the individual projects paid for the service. They had top level buy-in because the decision was made to load test all projects.

I was more interested in how Pfizer's process had to centralize, not just hardware and product licenses. I also asked Anant about the numbers of customers that are at a Performance Authority today. The number is still less than 1% and it's been that way since at least October 2004, when I spoke about the topic. Obviously companies aren't getting it. What do these companies need to do to reach that? I wasn't really clear on the answers. Greg feels that Pfizer is in a Service Utility and very close to the Performance Authority. I would like to see a presentation at HP Universe that talks more about how you know which stage you are at and how to get to the final stage. Hmmm. Would would I know who could submit such a presentation? Perhaps the one I submitted on that which got rejected. But I'm not bitter....%&%#%&!    :)

09:45 a.m. - As I was walking to my next session, Amjad Alhait (President of Betasoft) caught up with me to say hello. I was wondering if anyone was reading this stuff.  Thanks Amjad!

I was originally going to the session "Effective Quality Assurance for SOA" by McKesson but they were not really cool about being published. I understand, the speaker was just trying to make sure that he was in line with the corporate legal department. Been there. It's the reason I am a small business today. I really did not want to attend a session that I cannot comment on, as I am always looking for something to poke fun of as it makes me feel important.... :)

So...I went over to the session titled, "Performance Validation and Optimization of large-scale custom enterprise applications" by Deloitte Consulting. As an employee of Deloitte Services LP from 1997 to 2004, I looked forward to meeting some alumni of "Uncle D" as I call it.

Phil's Guiding Prinicples boiled down to "quality must be built in, not tested in". Everything he said was right on the money, but it did have the typical Deloitte buzzwords mixed in.

My only pet peeve, and this is not really pointed towards this presentation, but more of a general observation from me just being a loud mouth, is the terms "Verification and Validation" service (V&V for short). The ONLY one who can VALIDATE your solution or your product is the CUSTOMER. Development, Project teams, QA, etc... none of these people can validate anything. They can VERIFY it.

Verify = "Did we build the right thing?"

Validate = "Did we build the thing right?"

Building an application right means you met the requirements specified for the product. (Yes, I know "What requirements"...)

Your customer (the business) will let you know if you build the product correctly because they will tell you if it sucks, or they won't use it at all, or they will come up with another product that does the same thing.  

[Scott steps off of his soap box to pay attention to the presentation again]

Vinkesh started off by stating that planning was the most important things in performance testing and asking the questions "What is the objective? What are you trying to do?" Perfect! He talked about a HUGE government application that was highly customized, and how they approached testing this application. There were some good nuggets throughout the presentation and if you attended the conference but did not make it to this session, you should download the PPT if you can get it. It will be more informative for those of you with less experience with performance testing large projects, but the information is solid. It covers planning, script design, and execution.

10:15 a.m. - Raul DeLeon shares with me his picture from Tuesday night:

There is probably a story behind this, but what happens in Vegas...can only partially appear on this web site. There are least 12 geeks at this conference who now respect you, Raul.

11:30 a.m. - Back to the partner showcase area because that is where the food is. Got to see and hang out with the guys from CheckPoint Technologies, including Ken Arneson and Bob Crews. It's really cool that all of us from the old Mercury partner days still know each other and hang out, even though in some respects out in the field we compete with each other too.

Loren Peters (HP Sales Engineer for the SLED team) gave me an overview of BAC 7.5 with all the new integrations. The product has really matured over the years and the look of it is getting more polished with each new version.

02:00 p.m. - Caught a few minutes with Mark Tomlinson, product manager for LoadRunner. Mark used to work for Mercury, then Microsoft, and just returned to HP not that long ago. I have followed his career from afar, and he's been on the cutting edge of LoadRunner stuff for awhile. I remember seeing a screen capture video of him demonstrating the first Visual Studio add-in to kick of load tests of  C# code direction in the Controller. He can probably script in VUGEN with the best of us. 

02:15 - Caught the session "Proven methodologies for achieving Web 2.0 application scalability with HP LoadRunner and Performance Center"  

Some of the challenges they faces were:

1. the variety of communication channels
2. AJAX: UI activity and sever request are not fully synced
3. Business process flow modeling and transaction definitions
4. Maturity level of Web 2.0 technology
5. Load test was done during a new role out

HP followed up to talk about where they are going with VUGen and Web 2.0 protocol support. 

One of the questions brought up to HP was why AJAX C&S and other "Web 2.0" were not just additional functions to the existing web protocol. I heard their answer, but I also heard what was not said. I'll leave it at that.

03:45 p.m. - Attended the session, "Verizon Wireless : SynML and the HP LoadRunner Software Development Kit". The speakers on this included Siva Sivaganesh of Verizon, Faraz Syed and Phillip Mayhew - both with Genelogix. 

Erez Barak started the presentation off by stating that the world of software development is changing and LoadRunner needs to fit in this new world, even though it has been out in the market over 15 years. We're not going to develop a whole new set of products, and we're not going to go open source. At the same time, we want the development of LoadRunner to have some portion that is opened up to the community. This is where the Software Development Kit for LoadRunner comes into play.

Genelogix built a protocol based on a client need. Verizon created a product (Backup Assistant) that syned up from various devices like a cell phone and backed the data up to the server on the back end. The objective was to create something filled in the gap from what the client needed and what LoadRunner had available at the time.

Phillip Mayhew went through the process of how they created their own protocol. They used HEX Editors, packet sniffers, disassemblers, and custom tools. They captured the packets over the wire, and then used HEX editors and trimmed out the information that they did not need. They took the body of the message and used a custom tool that will put the XML message that was actually sent. The disassembler was used to figure out a session id that was being generator. Phillip had to look in a debugger and try to see what the id was. There was a WAP-Binary library that was available, so they did not have to worry about the translation. They took advantage of what they already had. They were able to use the existing web functions in loadrunner that were already in LoadRunner as well.

For those of you feeling froggy enough to create your own protocol, keep in mind that there are licensing restrictions on what you can and cannot do. You will want to consult the LoadRunner documentation on this and get with HP to understand all the limitations before jumping into this. It also helps if you are a masochist.

05:00 p.m. Heading up to the room to get ready for the conference party. Last chance to get incriminating photos. When I walked by the mainstage doors, some staff were walking out. I smell smoke. BBQ perhaps? If so, they might be sorry I am here... :)

For those of you who do not already know, I am a BBQ nut! Here is my blog on that:

www.bbqquest.com

06:30 p.m. - The doors open and we have a look around.

We head to the buffet (of course, you guys know me by now, right? Food is the MOST IMPORTANT part of any conference). I am allergic to shrimp (iodine) but I thought you guys might like to take a look at this. There were about 8 tables of this.

I can eat sushi, and they had some really good stuff available on the other side of that shrimp. They also had Korean style lettuce wraps, and some fajitas. The deserts included a mini-strawberry shortcake and chocalate cake. 

I got a chance to meet Jay Hakami of SKYIT Group, and extremely nice guy, with a company who is very partner friendly. In fact, all of the folks at SKYIT are top notch in my book. Lydia Consillas and I talked for just a few minutes and Tim and I got to meet Keith, their sales person in charge of requirements and their relationship with BluePrint. They have been doing some podcast and they have been very popular on iTunes, and I thought "What a great idea". Perhaps we will do a load testing podcast soon and see if anyone is interested in listening to what we have to say.

Some dude who got popular from some show on TV came out and sang and made noises with his mouth. Dude, I must be really getting old. By the time his set was done, I was reading the Drudge Report on my cell phone. 

I regretted not getting a photo of Eric Shumacher earlier, just because he is such a cool guy. So I finally found him and snapped this photo. 

We talked about our thoughts on the conference.

After this, I determined it was time to bail. So I missed the main act, whoever that was. 

THANKS HP, and thanks to everyone who made this conference possible. I know there was a lot of hard work that went on in the background and I appreciated it. My only complaint: get rid of those freaking mini-xylophone things that the staff walks around clanging on. I will be bringing a blunt instrument next year and there will be blood.

It was great to see old friends and make new ones and I hope you will all keep in touch going forward. I look forward to seeing you again next year.

I hope that whoever reads this gets a single person view of the HP conference and those that attended can read and enjoy it all over again through these updates. If you liked this coverage and want to have me do it again next year, please let me know. If you have additional photos, stories, or notes, please send them in and I will post them as well. 

Tim Chase has more notes to add from his sessions, and I will put them on their own page.

Take it easy...

Scott