InfoWorld CTO Forum 2001 | home
Thursday Morning
The morning panel was on "Managing Through a Crisis". (During the 7 am breakfast -- this conference went from very early to very late each day.) Moderated by Henri Asseily of Bizrate.com, it included Roger Burkhardt, senior VP & CTO of the New York Stock Exchange, Dawn Meyerriecks the CTO of the Defense Information Systems Agency, and Ronald Robinson the president of the technology division of Texaco.
The "Managing Through a Crisis" panel, NYSE's Roger Burkhardt being interviewed afterwards for TechTV: front and back
I really liked this panel. Again, very open, CTO to CTO talk. Ronald talked about the long-term crisis of having a new boss come in who listens to consultants and completely reorganizes things. Dawn and Roger gave more short-term "war stories". Dawn's actually was sort of a real war story. The U.S. government DISA mission is "To plan, engineer, develop, test, manage programs, acquire, implement, operate, and maintain information systems for C4I [Command, Control, Communications, Computers and Intelligence] and mission support under all conditions of peace and war". They are in charge of the U.S. President's "red" phone, etc. She got a call in the middle of the night. The question: "Hypothetically, could you hook up some particular communications system to some other?" She answered. A call a little later: "Hypothetically, what equipment would you need?" Soon, the next morning, the call: It wasn't hypothetical. She and some of her gray ponytailed and ear-ringed best were on a plane with some special forces-like guys. The military guys seem like they trust no one but their own. They're big -- "They look like they eat gravel for breakfast". And she had to ask them to give up their laptop computers to her guys (who looked, shall we say, quite different) to modify. That would be the only link they had to support them in a secret mission, dropped by parachute or something. No time for beta testing. Luckily, it worked.
Roger talked about the recent computer system outage at the New York Stock Exchange. They rarely have an outage that lasts much. Something like once every 10 years. They installed some new system. It didn't work, and, as always, they had a fall back to get to the previous system. Unfortunately, the falling back was the one case that exposed one of those special communications bugs that are hard to find. They have procedures for all this. They immediately checked what was affected. No institutional trading, just retail orders that were a small percentage of the volume. The head of the Exchange used principles they had developed in analyzing predictable crises and decided that the small investor needed to not be treated as second class, so they shut down trading. He told of working on the solution and looking up at the broadcast TV monitors showing his boss on live TV, who'd then excuse himself and stick his head around the corner from the camera and ask directly how it was going. Really in the thick of things.
Henri talked of a quick solution to a major last minute problem when they ended up using Unix GREP as their main search engine for a web site.
I love such stories. Like the days of case studies at business school.
For this I took notes on my Palm. Here it is on the breakfast table with another attendee using his RIM in the background:
Notetaking
Some of my notes:
Leadership vs. management: It's leadership, the CTO is like a coach.
Sticks neck out: wasn't looking for a job when got this one, so not afraid of finding another.
Decide in advance who will do what in crisis, especially who will talk to the press. "CNN determines what's a crisis"
Get experience with press, NYSE has 4 onsite studios so used to it but others need training.
Crisis gives momentum to get cooperation.
Q: How to do w/o crisis? A: Crisis helps change culture. Helped us "realize we are one entity".
Q: In crisis, what worse screw up? A: Not being assertive enough. Follow gut when things turn.
Q: What if can't fix fast, what PR? What post mortem? A: Would make it work :). Don't want crises but after having use them as a resource. Wished rest of team had been there to experience it to help build org to keep system live. The in gut feeling of terror will help build team. First time end up feeling in stomach, helps for next time - will help focus and keep from making mistakes again. Helps find leadership qualities in people you don't see until then, and who need help. Generates heroes (not goats). Reward them.
The next session was on "Managing Talent" with Michael Dunn, advisor to AOL Time Warner, as moderator, and panelists Frank Campagnoni of GE Marketplace Solutions, Gilbert Cloyd of Proctor & Gamble, and Timothy Villanueva of Hotjobs.com.
"Managing Talent" panel
It was followed by a panel on "Apps on Tap: Building the Virtual Global Entity". Moderated by Loudcloud's Tim Howes, it had Ejasent's Rajeev Bharadhwaj, vJungle's Phani Vaddadi, and Terra Lycos' Tim Wright:
"Apps on Tap" panel
Some raw notes from my Palm with very minor edits:
Loudcloud: capacity on demand - can scale on spike on demand. challenges include: ?
Ejasent: computing utility to run applications. How do you scale up and down? buy real estate all over the place - has bottlenecks, they now do micro-measuring so can charge and had to prove to ISVs it's OK.
vJungle:" Over the hump, but the hump seems to be getting longer and longer"; all apps talk XML to talk to each other (we were told he's the ex-Microsoft person who got them to do XML) -- enable passoff from app to app by XML. Problem was when no apps how to show will have problem when get there they would solve. Focusing on financial institutions. People don't trust data with startup, so do it private label. Problem at Microsoft w/ xml: alternative way to serialize code. Challenge to get programmers to do it. A way to transmit data outside of the firewall.
Lycos: buys apps on tap, ad serving, data, etc. [Dan's note: and Trellix Web Express...]: main benefit of outsourcing storage was s/w it came with, consumer/small business Telefonica needs usage based since consumers won't pay up front. Want to bill like pay-for-view. packaging up apps and then billing by usage. "Interesting business model". Looking how to diversify, corporate portals, web hosting, etc.: e.g. they did USGA for US Open and went from 0 to 40 servers on each coast, with 50 million page views. Will do "event like coverage" - Bertelsmann music events. It's about the Service Level Agreements (SLAs) and strength of vendors.
Q: Major challenges to move industry to this model?
Convincing ISVs that you can make money, including CFO. Need to have enough companies doing it. Must show them ROI and keeping control of customers.
ISVs etc. are scared of hooking customer to these vendors. Commissions are a problem -- need to make it like it was for ongoing sales - since don't have up front fee to get commission from. Accepting of having someone else manage some of your infrastructure is a tough thing. Deciding to choose SNI (Storage Networks Inc.) at Lycos was a long thing. Need good metrics -- tight SLAs need measures that are quantifiable.
Q: You all are infrastructure people. Where are the "content" people who will do the services other than web hosting? Who will be the actual providers that will show up in UDDI
A: Challenges being an ASP. Used to think all sw cos with migrate to ASP because good business model. Didn't happen.
Built this assuming asp's will come. If does then we connect them. Transitioning and building new ones like payroll, etc.
ASP model is they already have them and then connect (??) "as is application capture".
Runs at terra lycos outside apps they have in house (Peoplesoft, etc.) - "I don't care where it sits as long as I own the data." Doubleclick's problem is how to scale their infrastructure. QOS (Quality Of Service) has been challenged.
Q: Still struggling with definition -- if we build inside our firewalls we can build ourselves or buy from ISVs. Outside firewall: do I outsource whole thing or Best of Breed services and build the glue to put together. How do I get SLA/QOS, etc. Is that it?
A: Even if have Loudcloud host/do your stuff still have some services elsewhere. Used to be had to do everything yourselves, now you can get infrastructure from people like them. Future is build things out of services out on net.
I want to check my balance at one place, not worrying where each piece is. Will need infrastructure to do this.
Apps on infrastructure is old, now just apps, later will be services without worrying which apps. By outsourcing storage I don't need to know how to use EMC stuff. My ongoing carrying costs are lower because I can manage them better that lets me use more. If I don't like what Peoplesoft does, I own the data, I can change.
Q: hosting: going in house or moving. What about it?
A: People still believe outsourcing works, just had bad experiences...
Financial institutions have legal requirements, so better for them to host, but just better to use pay for use then other way, doesn't matter where host.
Lycos: outsourcing is better because cheaper, better peering, etc. Hosting will be outsourced because it's a commodity. Why would you want to own that on your balance sheet?
Eric Drexler on the big screen
Notes:
"Nano technology scares the heck out of me but that's because I understand it." Have had micro technologies for years (e.g., integrated circuits). Other fields, like bio technology, but it's natural nano technology. Interest is growing in nano tech. Senior people are ignoring, but happening. Molecular manufacturing will be the core. "Hands" that pick up things and put them there. Chemical engineering works like putting parts in a bin and shaking it up hoping things get together. Things you build will be different than bio, but same principles, like bird vs plane (still uses lift) or brain vs computer. Non-molecular things we manufacture now have defects and hope work OK, molecular will not have that - nano tech is about building perfect things at the atom level, like digital technology is "perfect" if not much noise. You get small scale phenomena and fast. Perfect, fast, precise and reliable (except for sw which is always messed up...) 1 cubic nanometer can hold 100 atoms, could be a switch. If tackle Babbage type machine down to nanometers with carbon-based stuff, could be tiny (1 cubic micrometer) and very powerful. Examples of use: fighting disease by attacking and dealing with aging. Body is nano molecular tech, so logical to use that to work on it.
Biology shows that molecular manufacturing is a good way to go. Cost per Kg of making something (after IP cost...) is low. Eg., if used best known material to make space shuttle, it would be much lighter. SUV made out of that would be able to get to orbit. Why aren't solar panels cheap? If can make better materials with atomic precision and cheaper, so "oil in the ground is grossly over valued" so price will drop and we'll use up what we have. Used to be if you wanted the sound of a symphony in your home you needed all the orchestra and the space for it. Now how have device you wear on your hip and have universal sound generating device. Used to use metal to create printed material. Now we have a "universal dot making device". Well once upon a time if you wanted to make a complex artifact you needed lots of people, equip and capital. In the future there will be a "universal atom placing machine in 3D". Can make itself.
Why look ahead? Well it's coming. Could be used for bad. Profoundly disruptive technology. And will arrive in our careers.
Quite cool.
|