Straight talk on Web services
E-speak co-inventor and Confluent CEO Rajiv Gupta has some hard words about the folly of trying to shove raw technology--no matter how leading-edge--down the throats of customers.
The co-creator of Hewlett Packard's ill-fated E-speak, one of the earliest versions of what would become known as Web services, Gupta lived through the frustration of being early to market with a new--even grandiose--computing vision.
With the passage of time, information technology shops have come to have a far clearer idea about what Web services are. But when it comes to convincing corporate chief information officers what it can do for their businesses, Gupta, now CEO of Web services start-up Confluent, believes technologists continue to commit a cardinal mistake.
"We should be selling Web services within the context of what the customer needs," he says. "But we're becoming too technology-specific and enamored of the technology."
That's become even more of an issue at a time when chief financial officers are reluctant to sign off on "iffy" big technology implementations, he added. CNET News.com talked recently with Gupta, who will officially unveil Confluent to the public later this month.
Q: You were the co-inventor and general manager of Hewlett-Packard's E-speak. I guess one might argue that you guys were ahead of the curve in Web services. But why did E-speak fizzle?
A: It was my baby. There were a number of different reasons for its failure; I made a bunch of mistakes, HP made a bunch of mistakes. And as you said, we were early--especially in the technology. Sometimes it's more of a curse to come out early than to come out late.
My mistake: I focused more on the technology than on the business context in which the technology would be applied. It's a classic mistake that people make, and I fell into the same trap. I'm seeing that (mistake) being made by others today. So, for example, we focused on the architecture being very plain and rich, but didn't focus as much on usability. We focused on changing the world, but focused less on helping the world see the value. We didn't appreciate the fact that the world wanted to change not as fast as we wanted to move.
Was it a curse to be out ahead of everybody?
Much too much is made of first-mover advantage. It's probably as much a disadvantage as an advantage.
| Sometimes it's more of a curse to come out early than to come out late. |
HP didn't really appreciate what it takes to set an industry standard, especially in software. It was a market where we didn't have the credibility that HP has in printers. So we very naively put our software out there as open source.
We thought that was sufficient but we didn't pay enough attention to building consensus and partnerships and driving those standards. Still, the vision and strategy and technology were right on. Now I just have to make sure that we don't make the same mistake again.
Was it something to do with the pre-Carly Fiorina culture at HP?
Some of it, yes. We were very much like a box company--whether it was boxes that spat out dots on paper or blinking lights.
What's to prevent Web services from suffering a similar fate? If it's not helping the return on their investments, IT managers don't want to hear about big new expensive projects.
Precisely. It's the people who have the burned fingers who, having made mistakes before, won't make them again. With E-speak, we didn't pay much attention to standards but today, all the players are paying homage to the same set of standards.
The downside is that many players are not paying as much attention to the business context and singing the praises of the technology. That has to change.
| People are not as willing to open their wallets, because they have technology indigestion and they are not as willing to listen to a new technology story. |
It's being sold as a panacea, and that is a mistake because you don't want (users) to get cynical too early. There is a place for Web services...where they can and should be applied. We should be selling Web services within the context of what the customer needs. But we're becoming too technology-specific and enamored of the technology.
Are there still problems with scalability, as Web services gains acceptance on enterprise deployments?
No, I think that is an implementation maturity. Does CORBA (Common Object Request Broker Architecture) have transactional maturity? It's how you implement it. The technology by itself doesn't hurt or help you. The same is true for Web services.
How about security?
It is an issue. To me, it's more a security-management issue than security itself. It's how to manage security policy--that is the bigger issue. It's more to do with visibility and control across the board.
Where does the technology need to go next? Is it a question of setting standards or some other kind of hurdle?
Yes, there's a need for standards, and some of the standards are in the process of being adopted and defined. But there is enough standards and technology out there already for us to do serious deployments. I think there's just a maturity issue.
The economy is in a funk and spending--well, it ain't 1999. Is that going to slow the proliferation of Web services?
Overall, you're right. People are not as willing to open their wallets, because they have technology indigestion and they are not as willing to listen to a new technology story. In this environment, you don't need to hook up to 2,000 dot-coms. But for companies that understand they need to get operational efficiency of existing investments, this could be very good for them.
Confluent is launching at the end of this month. You really picked one heck of a time to start a new company.
This is either the best time to start a new company, or it could be the worst time.
You were one of lead architects for Itanium. Why did it take so long for the project to get completed, and why do you think it's not selling?
When I first started work on the HP project, my intent was to kill Intel, not partner with Intel. At the time, (Itanium) was to come out in 1993. Obviously, it has taken much longer than originally imagined. With Itanium, too many things moved at the same time--a new architecture, new compilers, etc.--so that had its effect on the delay.
By the way, I think working with Intel was the right thing to do. I spent a lot of time bringing Intel up to speed--there were a number of issues.
It will take time (for Itanium to succeed). The unfortunate part is that in this economy, time is not a friend. But in terms of its raw capability, I think over time it will be a success. The new (Itanium) already is much faster than any other processor on the market. Now the question is how valuable that additional speed will be in this economy.
