Saturday Jul 22, 2006

The Long Tail of Web Applications

Listening to National Public Radio (NPR) the other night, I was intrigued by an author, Chris Anderson who coined the term “the long tail” of sales in his book: The Long Tail: Why the Future of Business is Selling Less of More. Chris used an arbitrary online music (by the song) retailer, and quoted something like 95% of the inventory of the eTailer has sold at least one copy, and something like 85% of the revenue is from the things that represent the bottom 85% of the inventory... this means that the OLD 80:20 rule (80% of revenue addressed by the to 20% of products) may now be proven wrong by the channel dis-intermediation offered by the Internet.

Translating this into something that I'm passionate about, Utility Computing, may mean that the revenue that utility providers should be chasing might not come from the “block buster hits” = the applications that people fund having built today, but rather from the 80% of applications not yet invented because the communities that they would benefit are so small that today's assembly techniques prevent their development (from a revenue/margin justified business position). Today, because of Utility Computing providers (pay-as-you-drink), productivity enhancing languages like Ruby, and dynamic application assembly techniques (“mashup” seems to be the word de vogue), perhaps UC should be banking on some of the enabling technologies to broaden the field of available applications... something unique for everyone vs. one size fits all.

Thursday May 25, 2006

The Sun Grid $50k Contest & the Doubter

While I appreciate Greg Nawrocki's perspective on the challenges associated with growing a community of users for Sun Grid and more specifically the Cool Apps Challenge, and respect his viewpoint; I do believe that there are a couple of things that Greg needs to remember.

  1. Grid computing has been around for a very long time, mostly in the academic spaces, but also in enterprises.
  2. The Academic / Scientific endeavors tend to frequently push the bloody edge of technology, which means that they are constantly battling for more “resources” to solve their mathematical/relational problems.... Chips are “never fast enough”, “memory is never fast enough or big enough”, “storage...” you get the picture.
  3. Enterprises are dealing with substantially more data than they ever have, and the analysis of that data - turning data into an actionable thesis is seen even more strongly as a competitive imperative.... this is taking enterprises into the theoretical realms once reserved for science.
  4. At the most basic level Grids are nothing more than a collection of resources, that can be combined to produce a variety of systems, though Grids today address the more data-crunching intensive workloads - some embarassingly parallel, there is nothing preventing a grid of resources from being used to compose an enterprise system substrate, in fact Sun's N1 initiative, Platform's Symphony, Cassatt Collage, DataSynapse, Parameus, GigaSpaces and others share this vision of composing systems from common components (should sound familiar to the SOA crowd).

All in all, Greg is right, having a competition is a mechanism to increase awareness of the Grid's capabilities and applicability but, I do have to ask “why not.” If the killer application for the grid emerges, great, if it doesn't; it has still increased the visibility of the value of utility computing and the impact of a mesh of consistent resources. Are developer contests right for the enterprise? - hey, who knows. Because Southwest Airlines emails out a “Cities Special” do you refuse to fly? And, what about Google's Summer of Code, or the Intel Game Demo Contest let's face it competitions are another form of promotion, and as an effective brand and community building technique to the all important developer community - independent or enterprise (most of us do both anyways).

Most enterprises share a set of challenges; now that we have already seen the impact of platform consolidation on both acquisition costs, and operations costs, and now we're just beginning to address the impact of virtualization on data center utilization (raises it) and operational factors (makes it more complex to manage). Virtualization impacts every layer in an enterprise architecture, and applications need to be better able to deal with horizontal scale/distributed workflows, and the impact of accumulated MTBF. I talk frequently about Deutsch's fallacies; distributed partial failure needs to be addressed explicitly by an architecture, and increasing awareness in this area is important for the entire IT industry. Sun is actively building a community of partners and customers who, in a like minded way want to solve for increased operational / capital efficiency, increased scale, and increased agility - difficult problems, but things that only a large community of constituents can solve.

So Greg, we had to start somewhere! if you know of a better way to get to our end goal of agile-distributed-component based-failure resistant-capacity managed-systems, I'd love to hear it!

Tuesday Jun 14, 2005

OpenSolaris “More Open than Open”

Rob Gingell (a former Sun Fellow and Chief Engineer) once asked me to help put together a white paper on how Sun was “more open than open” which is to say that there are so many companies out there who touted Open Systems, Open Software and Open Standards, for whom their motives weren't on the straight and narrow. I wish that I had had this entry.

“Today we have taken our crown jewels, our flagship piece of software, Solaris and releasing it to the Internet community

I'm most excited about “what's in there” specifically today's consolidation is the core “OS/Networking” what we're calling our “ON” gate release, which as I have been playing with for a couple of weeks, delivers a very stable environment for most ISP like functions, and is the base for all of the other bundles / packages that “live on top”. And the ancillary developer tools - both Sun's Studio Tools and those from GNU.

Like Red Hat is to Fedora, OpenSolaris provides a community development pool, and Solaris becomes Sun's distribution. For the past months Sun has been seeding this community with pretty large tarballs (~ ) which could be manually built, today we're launching an interactive site, typical to most development programs, the CVS tree will be a managed entity.

And since I know that there are going to be a lot of takers, sorry if the download is a little slow - opening day is always popular.

Have fun ;)

Thursday May 05, 2005

The Old IT Is Dead. Long Live the New

I've been sitting on an article for a little while (4/18). I just found it refreshing that a journalist finally said something that made sense in explaining today's corporate buying & investor strategies.

“the industry is on the cusp of a sweeping change to new information technologies such as true mainframes-on-a-chip, Web services, and open-source software... Everything we talked about in the '70s, '80s, and '90s -- putting together clusters of PCs to replace big machines -- is finally happening.”

And finally a statement that I fully support...

“The fact that the next generation is less expensive does not mean that growth disappears. If you wind up uncovering significant new ranges of applications and you end up deploying them far more widely, you're going to dramatically expand digital services.”
I think that most everyone recognizes the growth of stored data that everything from compliance to RFID are driving to record levels, and it can only be expected that people are going to want to begin to capitalize on this expense that they are incurring, through programs including statistical analysis for business process optimization, product quality / time dependent reliability, or financial planning. What was holding us back? I can only suggest that it had something to do with cost, something with scale, and something with complexity. Only by looking systematically across these factors can we finally realize a solution.

Thursday Apr 21, 2005

Application Containers & Cost Per Transaction

Obviously critical to Sun Grid's success is the ability to support transactional applications on the compute farm. Though the core economics of a Solaris 10 cpu at $1/cpu-hr are best in class (versus $7-$14/cpu-hr for many self hosted enterprise environments), who would have thought that the economics could get even better so quickly. Today, Sun announced couple that with some recent performance benchmark figures from our Java Application Platform:

v20z topview

“The benchmark configuration was comprised of 13 Sun Fire V20z servers, each equipped with two AMD Opteron processors Model 248 and running the Sun Java System Application Server 8.1 Standard Edition. In addition, a Sun Fire E2900 server, equipped with 12 UltraSPARC IV 1.35 GHz processors, ran the latest version of the Oracle 10g Enterprise Edition Database.

The benchmark delivered a class leading price / performance of $164/JOPS in the application tier (combined software / hardware and support cost) versus IBM's best submission of $504/JOPS(3). This translates to 90% of the performance for 33% of the total application tier cost.

The benchmark also demonstrated Java System Application Server's ability to handle complex workloads under very heavy load. During the benchmark, the Application Server supported over 9400 concurrent clients, and processed over 4.3 million complex transactions per hour.”

Wow! combine this with the reduced cost and complexity associated with a unified development, test to deployment model in which the integrity of the “digital runbook” is maintained, and people should start to ask not IF but WHEN refactoring for Sun Grid should occur.

Keywords: , , ,

Wednesday Dec 29, 2004

Happy Holidays

Happy Holidays to everyone.

We're on a 1 week break between Christmas and New Years. I'm taking this opportunity to explore a variety of Blog clients including ecto and Mars Edit for my Mac. I'm very interested in scripting languages, so let's see what these graphical front ends can do.

Wednesday Nov 17, 2004

Immersion Week

We've just wrapped up Immersion Week for our Client Solutions Team, extended with a number of partners that make up our “infinite bench”. Have to say that I am really pumped up, after the Launch on monday, the fantastic deep dive sessions on Solaris 10 and our practice disciplines.

Just wanted to recount one thing that I heard that was cool: “I know that you've caught a lot of flak over the mandate to install Solaris 10 on your laptop before attending, but I'm seeing that 80% of the people not only are happy that they did so, but are now actively working in that environment.”

To me this just puts an exclamation point on the notion that Solaris 10 is really cool. If you're a geek, it's got more capabilities than you know what to do with - like a kid on Christmas. And if you're not a geek, the differentiated value is so obvious:

  • 30% performance increase in the network stack from a simple install - wow we just made your hardware that much cheaper
  • a thousand zones (Solaris Containers) on a single CPU, allowing unfortold benefits from consolidation - cheaper again
    Fault Management Architecture(FMA) and Systems Management Framework (SMF) - just keeps running and running, cheaper
  • OpenSource and free RTU - cheaper and cheaper

IMO this is a quantum leap forward, go get your copy today from Solaris Express.

Friday Jul 30, 2004

System Integrators are Great

I just returned from a couple day retreat with key SMI partners. I have to say that I am heartened to hear them say that our new Client Services organizational structure is just what they need to embed our distruptive technologies, and key differentiators into their programs. As many may know, we are actively building tremendous depth in key areas from our Desktop & Mobility space, our Enterprise Web Services program, and others.

Also very cool is the SunLabs open house. Though I didn't have time to attend this week -due to this SI meeting in Dana Point, I did review the curricula and it's amazing... the coolest is Mackinac, our Real Time Java VM implementation (JSR-1), which really sets the bar high. For enterprises that are talking about "real time" transactionality, this give us the application substrate.




« July 2014