Web Services Strategies
What
Web Services Will and Won't Do. Three Gartner guys "get
it," and cut through some of the hype. I particularly appreciate
that they dispell the myth that, "web services will foster a revolution
in self-aware software that makes all decisions without human
intervention."
Posted Friday, March 15, 2002 6:01:31
PM
|
|
Ecademy.
Thanks to Julian Bond for
turning me on to this UK-based site where web services are a major
focus. It takes a while to find your way around, but there are
numerous treasures here. Thanks to David
Rosam, too, for recommending the site.
Posted Friday, March 15, 2002 2:36:05
PM
|
|
IT Strategy Letter.
If you don't have the ability to subscribe to my weblogs via RSS
(using a tool like Radio),
you can now receive the same content via email by subscribing
to my new newsletter.
I expect to send it out weekly. Either way--RSS or email--sure
beats having to visit the site just to see if anything's new.
Posted Thursday, March 14, 2002 12:03:27
AM
|
|
Tale
of two Rodneys. Steve Gillmor at InfoWorld deciphers the
politics of the Web Services Interoperability (WS-I) Organization
of which Sun is most notably not a member.
Posted Tuesday, March 12, 2002 11:58:42
AM
|
|
Catching Up on My Reading. There's
so much being written about web services, I'm convinced you could
read for 40 hours a week just trying to keep up. But I've managed
to get through a huge pile over the past few days. Some of the
best writing I've encountered comes from The Stencil Group, which
I mentioned
a few days ago. One favorite issue is by Brent Sleeper from July
15, 2001., and not just because I used to play Mille Bornes.
Not part of the series, and harder to find on their web site,
is a lengthier 12-page whitepaper written by Bill Robbins, entitled
How
Web Services Wil Beat the "New New Thing" Rap. Some first-rate
thinking and analysis there.
Another great resource is the Web
Services Zone of IBM's developerWorks site. Catalysts
for fee-based Web services is the latest (and best) in a series
by Dan Gisolfi, an excellent writer.
Posted Saturday, March 09, 2002 5:54:28
PM
|
|
Reality from the Front Lines.
A friend and client, in the middle of a web services implementation,
has this to say (edited):
We're in the middle of SOAP, and at this point, I'll
vote "not ready for prime time," at least on the Java side. XML
is just a huge mess. JAXP, Xalan, Xerces, Crimson, SOAP, Xmlparserv2.
Everything requires something else; this can't work with that;
you must have version x.x.x of A to work with B, but of course,
that won't work if you're using C. The general consensus [among
deeply involved developers] is that SOAP is a technology to watch,
but not many people are willing to tackle it yet. This is one
instance where having a single entity like Microsoft around to
put together a logical and functional set of parts would make
for a better and more effective developer experience. [On the
non-Microsoft side] we've got Sun, Apache, IBM, Oracle and others
all trying to come up with the right "solution." But right now,
it's a mess of interoperability. Sun is trying to address this
very problem with their XML bundle and Java Web Services Pack.
Let's hope it works.
Posted Saturday, March 09, 2002 9:30:54
AM |
|
<E-Business*Standards
Weekly Wire/> Weekly Wire. Yes, the title includes
the XML angle brackets and slash. (What a pain to make it render
in HTML. :-)) E-Business*Standards*Today, published by Data Interchange
Standards Association (DISA), provides a daily update on the latest
developments in e-business standards, from a variety of news sources.
It's a good way to learn about progress on industry-specific XML
standards. Free subscription form at the bottom of the page.
Posted Friday, March 08, 2002 9:24:01
PM
|
|
Frictionless Discovery. According
to the hype, UDDI directory services will allow your systems to
automatically switch from the web services of one vendor to another.
Have you studied UDDI? It doesn't come close to supporting such
flexibility. As Brent Sleeper at The Stencil Group wrote
back in July, 2001, "Although much of the published enthusiasm
for web services seems focused on a utopia of frictionless discovery
of trading partners and minimal switching costs among programmatic
connections, corporate IT users know this vision is a long, long
way off."
Intel's Chief e-strategist, Christopher
Thomas believes this "second step of web services is three
to four [years off]." It's one thing to discover that a service
exists (even that remains a manual process with UDDI), but unless
all of the competing service providers adopt identical SOAP interfaces,
you've still got to code for each one. Perhaps this will be an
interim
role for web service networks like Grand
Central and Flamenco--providing
a single interface to comparable but non-identical web services.
Posted Friday, March 08, 2002 9:16:32
AM
|
|
Live
on CNET Radio. You can catch me live on CNET Radio on
Wednesday 3/13/02 at 10am PST (1pm EST). I'll be talking about
web hosting and answering questions from callers for a full hour.
If you're in the S.F. Bay Area, you can receive the show on 910AM.
In Boston, it's on 890AM. Otherwise, you can hear the show live
on the Internet.
Posted Thursday, March 07, 2002 4:42:57
PM
|
|
The Stencil Group.
Created by former iXL strategists Brent Sleeper and Bill Robbins,
this consultancy publishes an excellent
free newsletter. Very valuable for staying abreast of developments
and ideas in web services. The back issues are worth downloading.
Posted Thursday, March 07, 2002 7:10:26
AM
|
|
Suddenly Seeking SLAs. Today
we use software components. Tomorrow we'll use Web Services. Today
we depend on warranties and support contracts. Tomorrow we're
going to need service level agreements. If the SLAs for Web Services
are anything like today's SLAs for connectivity and web hosting,
we're in trouble.
I see two problems. First, the typical SLA in entirely insufficient
as an expression of an acceptable business relationship. It's
a long story; buy my book
and read Chapter 10. Second, as mentioned in the Icebergs Ahead!
posting below, we're going to be receiving our services from vendors
who have never done this before.
I predict the emergence of a new business: third-party monitoring
of Web Service service levels. Companies like Keynote could play
this role, but I think it's more likely to come from new players.
Posted Monday, March 04, 2002 7:27:52
PM
|
|
Icebergs Ahead! The biggest
challenges of Web Services facing CIOs and CTOs won't be the technology;
it's fairly simple. But the migration from software components
to Web Services brings with it some implications that few have
yet considered. For example: today, your B2C e-commerce application
computes sales taxes using code/data modules licensed from a third
party. Your tax calculation process is relatively stable. Once
it's working, it tends not to break. The worst thing that happens
is that updates don't get to you on time or they contain errors,
but these are relatively simple to fix and the system doesn't
tent to work/fail/work on a minute-to-minute basis.
In the Web Services world, however, you'll compute sales taxes
by making real-time requests over the wire. Your sales tax vendor
will then have a whole new set of challenges. Does the vendor
(today a supplier of software and data) have the experience, staff
and infrastructure necessary to keep a Web Service up and running
24x7? Most software companies don't know how to do this; they
only know how to write, deliver and maintain code. Suddenly, a
feature that broke only occasionally is likely to fail more frequently
even if for shorter periods of time. In Web Services, the MTTR
increases, but the MTBF drops dramatically.
Furthermore, if your e-commerce application requires multiple
Web Services to complete the checkout process, your likelihood
of failure increases for each service upon which you depend. To
mitigate these increased failure rates, you'll need to plan to
utilize multiple tax calculation Web Services, in turn
supported by your own failover logic. Hmmm...this is starting
to sound a lot harder than the way it works today! You're going
to be planning for Web Service single points of failure just as
you do today for hardware.
And this is just the tip of the iceberg. If you intend to be
a Web Service Provider yourself, are you up to that task? Are
you prepared to deliver your service to meet high-availability
requirements? Do you have a track record of doing so?
Posted Monday, March 04, 2002 6:20:24
PM
|
|
Web Hosting Strategies
CarrierHotels.com.
A message from Rich
Miller reminded me about this site that I'd found a while
ago. It's aimed at the vendors, helping them to buy/sell/lease
colo and telco space, but if you're a high-end customer, you'll
probably find some gems here. At $1,795 their Colocation
2002 "450-page guide to the industry of power and space" is
a bit pricey, but looks interesting if you're spending more than
$20,000/month for pure colo.
Posted Friday, March 15, 2002 3:45:53
PM
|
|
IT Strategy Letter.
If you don't have the ability to subscribe to my weblogs via RSS
(using a tool like Radio),
you can now receive the same content via email by subscribing
to my new newsletter.
I expect to send it out weekly. Either way--RSS or email--sure
beats having to visit the site just to see if anything's new.
Posted Thursday, March 14, 2002 12:03:27
AM
|
|
Wimpy SLAs. If you've read Chapter
10 of my new book, you
know how ridiculous I think the current state of affairs in SLAs
is. On behalf of a client, I recently had occassion to review
WorldCom's SLAs for shared
hosting and managed
hosting. I don't want to pick on WorldCom. At least they put
their SLAs on line for all to see. And their SLAs are no worse
than most others I read. But SLAs like WorldCom's are just about
worthless. Just some pet peeves:
- Scheduled Maintenance. The shared-hosting SLA includes
28 hours per week of potential scheduled maintenance time during
which you can't complain and have no rights in case your web
site is down. IOW, 17% of your uptime is up for grabs.
- Outages. Again in the shared-hosting SLA, a problem
isn't considered an outage unless it lasts at least one hour.
Your server could be re-booted every 20 minutes, and take another
ten just to come up. 50% downtime, and it's all "okay."
- Remedies. And if your server is unavailable for the
entire month, what do you get? You'll get a credit for one entire
week, thank you very much. And you have to ask for the credit--it's
not automatic. And you won't get a credit for the bandwidth
you paid for and didn't use while your server was down.
I could go on and on, and I do in my book, but the fact is that
these SLAs are a joke. And like I say, it's not just WorldCom--they're
all like this. Think I'm nuts? Imagine if you bought a new car and
the warranty read something like this: "We guarantee that this car
will run forever, but if it doesn't, we'll fix it for free for the
first five years or 50,000 miles whichever comes first. That is,
except for a few minor inconveniences: (1) With advance notice,
we may disable it for up to 17% of the time. But don't worry. That
will only happen when you don't need it very much. (2) If your car
turns out to be a lemon, and we can't keep it running, we'll give
you back an entire 25% of what you paid for it! (Fine print: We'll
also have to charge you a fee for the miles you weren't able to
drive while your car was out of service."
Think I'm nuts? Ask any web-hosting vendor and they'll tell
you it would never happen like that. Their SLAs are far more conservative
than their actual performance history. Oh yeah? Well why don't
they make their SLAs more closely resemble the claims they make
during the sales cycle? My car's warranty exactly matches the
service I expect from the dealer. No more; no less. Why isn't
this the case for SLAs?
Posted Saturday, March 09, 2002 5:12:15
PM
|
|
Stats
and Trends. There are no new concepts in the PDF
whitepaper written by dash30 contributer Gilda Raczkowski
and published by WorldCom, but it's a free source of interesting
stats from Forrester Research. The numbers include web-site, hosting
and e-commerce trands by company size, and growth by type of hosting
service (shared, colo, managed).
Posted Friday, March 08, 2002 8:41:00
AM
|
|
Live
on CNET Radio. You can catch me live on CNET Radio on
Wednesday 3/13/02 at 10am PST (1pm EST). I'll be talking about
web hosting and answering questions from callers for a full hour.
If you're in the S.F. Bay Area, you can receive the show on 910AM.
In Boston, it's on 890AM. Otherwise, you can hear the show live
on the Internet.
Posted Thursday, March 07, 2002 4:42:57
PM
|
|
Verio
Closing More Data Centers. "Verio announced a restructuring
in September to trim staff and close some of its 46 data centers
as it switched its focus to providing managed services. At that
time, Verio did not say how many data centers it would close,
but a spokesman says about a dozen data centers already have been
shut down. Last month, Verio revealed that it is making deeper
cuts and will close a total of 36 data centers, leaving 10 operational."
I don't normally report news, and I have no inside information
on this, but some of my clients are hosted at Verio, and they've
been satisfied. It looks like Verio is biting the bullet and doing
what they need to do.
Posted Sunday, March 03, 2002 10:05:16
PM
|
|
Subscription
and Contact Info
The IT Strategy Letter is published weekly by Doug Kaye.
The content is identical to Doug's
weblogs.
|
|
|