Wikipedia doesn’t seem to know just when the term “application server” was coined nor who coined it, but if “information server” becomes popular, it’s a safe bet that IBM will get the lion’s share of the credit. That, to my eye, is the big news out of IBM’s Information On Demand conference. True, there was mention of the completion of the FileNet aquisition, some improvements to their Master Data Management (MDM) offerings, and some new vertical specific frameworks, but the Information Server was the primary topic of conversation today. Almost to a fault.
The question on everyone’s minds, and likely some of yours, was simple: what’s an information server? Let’s take a look at that with that a good old Q&A.
Q: Before we continue, let’s get the disclaimers out of the way.
A: Sure. IBM, as most of you know, is a RedMonk client. Further, both James and myself argued on behalf of the information server concept since first learning of it quite some time ago, so we’ve been advocates for it in the past internally at IBM. We have not been involved in its recent product strategy, however. Oh, and IBM comped travel to this conference.
Q: Concise definitions of just what an information server is are appear to be in short supply. Is it just a glorified database?
A: No, it’s considerably more complicated than that. The IBM folks have been comparing it to the application server, and the comparison is apt – for better and for worse. This wasn’t precisely how it was drawn up for me, so take the following with a grain of salt, but I’d say that the application can be considered in three logical tiers: a platform foundation, a layer of information oriented services that sits on top of that, and a series of optional components. In a bit more detail, it looks (something) like this:
- Optional Components:
- Understand (analytics, roughly translated)
- Cleanse (normalization, etc)
- Transform (yes, the “T” from ETL)
- Deliver (federation capabilities, otherwise known as EII)
- Services Layer:
- Parallel Processing
- Platform Foundation:
- Application Server (WAS currently, support for BEA and WAS CE planned)
- Relational Database (DB2 by default, will support Oracle @ launch)
So as you can see, the Information Server includes a database, but a lot of other componentry as well.
Q: Well, there’s a lot to parse in that explanation, but let’s leave that task for a subsequent Q&A, and move on to functionality: just what is this thing supposed to do?
A: IBM’s press release says that it’s designed to “deliver trusted, consistent and reusable information to applications and business processes.” Press release-speak, I know. In English, I’d say that much as the application server was intended to be the platform for the delivery of…well, applications, the information server is intended serve up just that.
Consider the trends affecting data today: hypergrowth of data volumes, increasingly disparate formats (including both structured and unstructured), along with increasing desires to deliver information horizontally rather than vertically to a variety of applications. The product is designed with those needs in mind. As I put it to one attendee today, the product is almost an EII product on steroids; it can collect data from a dozens of different sources, clean it up, and provide centralized querying or delivery of said data depending on the requirements. Additionally, according to informal feedback from the 75 or so beta customers, it’ll scale while doing that and be relatively straightforward to administer.
Q: What about logistics: when does it arrive, how much does it cost, and so on?
A: The Windows version will arrive first, at the end of November, while the AIX, Red Hat, Solaris, etc versions will arrive shortly after that – probably in the December timeframe. Costwise, it’s not cheap: starts at $100K, and that’s stripped of most of the optional functionality. As you add in functionality, it gets significantly more expensive.
Q: How does the price strike you? Will the market bear that cost?
A: Depends on which segment of the market we’re talking about. The high end crowd will almost certainly certainly invest. If the product can deliver on its promises, I’m sure IBM will be able to sell this to a significant portion of their higher end customer base, because if you’re the type of customer dropping $250M on a datacenter the cost is incidental.
The problem is that there are only so many customers with that kind of budget. At the current time, the product is likely to be completely out of reach of the small business, almost entirely out of the reach of medium businesses, and problematic for some of the smaller large businesses. A volume product, this is not. And that’s unfortunate.
Q: Why unfortunate – shouldn’t IBM maximize the margin on newer products such as this one?
A: Perhaps. The difficulty is that the problems the Information Server is designed to address are not unique to larger businesses, any more than the tasks addressed by the application server are. There are undoubtedly countless upper medium sized businesses that would benefit from such a product, but be unable to afford it.
Q: Are you implying that smaller and medium sized businesses are somehow entitled to the product?
A: Not at all. IBM certainly has the right – not to mention a mandate from shareholders – to monetize its product as best it can. What I’m questioning is whether or not targeting a higher margin, lower volume market such as the one currently open to the Information Server is the best economic returns. In IBM’s defense, however, this may be a strategy dictated at least in part by technology.
Q: What do you mean by that?
A: Well, consider the first iterations of application server technology. They were big, complex and required some fairly beefy hardware to run on. I should know, I was on a lot of those buying decisions during the bubble. From what I gather, the Information Server is at a similar point in its lifecycle; it scales, and it performs, but is quite demanding on hardware. To the point that it might overstress the limited IT infrastructures of smaller customers. Hopefully, over time we’ll see the product evolve and the business model along with it, and even small customers can begin using information servers – where (if) it makes sense for them.
Q: Why if?
A: Well, as I put it to eWeek, I think the requirements for database environments today are multi-faceted. MySQL, for example, is quite content going after business that’s an afterthought for the likes of IBM and Oracle. A big part of their success in those markets, as I’ve argued before, is because of their simplicity. My DBA skills even back in my development days was limited to a rudimentary understanding of SQL, and an advanced understanding of how to drive DBAs crazy with inefficient queries. Despite that, I use MySQL today within RedMonk’s infrastructure with ease; it’s just not that hard to use.
At the higher end, however, you have customers demanding quite a lot of their IT environments; one of the cited customers during yesterday’s analyst Q&A is running the Information Server beta on 6,000 nodes. Those types of customers, frankly, are those for whom “good enough” is never good enough.
From what I’ve heard to date, the Information Server is going to be a real find for the latter type of customer, and of limited interest to the former. It could be sigificantly interesting to smaller customers, but would have to become a bit more like Tomcat and a bit less like WebSphere XD. So that’s an “if” at the current time.
Q: Is this a response, do you think, to some of the customer frustrations described in Adam Bosworth’s ground-breaking “Where have all the good databases gone?” entry?
A: As an aside, I sincerely miss hearing from Adam (though hearing from his son Alex is similarly rewarding). But the answer is: not really. The federation portion of the equation would seem to address some of the problems of dynamic schemas and so on – at least in part – but overall I see this product as responding to a different sort of need. Part of that mission, clearly, is services enablement – information as a service, as IBM would put it (which may remind some of you long term RedMonk readers of COA) – and a bit portion of it is piece part consolidation (or integration, if you prefer).
Q: Are you a believer in the information as a service concept?
A: Just read our paper on COA. Our belief is that making information available horizontally is ideal not just for compliance challenges, but for application architectures more generally.
Q: Do you think that the term “information server” will become as commonplace as its application counter part?
A: I asked just that question yesterday in the press Q&A, and the answers I got essentially amounted to tough to say, but maybe. Certainly you’ll see IBM partners line up behind the term and begin using it; the launch list of supporters for the technology was impressive indeed. But it would only become commonplace if either a.) IBM’s competitors – notably Oracle – began using it, or b.) customers took it mainstream. I have doubts about the former, because Oracle doesn’t want to let IBM frame the technology discussion after leading the RDBMS mindshare game for so long, and the latter because there are some educational challenges to the term as described above. So short term, I don’t expect the term to enter the technology vernacular, much as I like the term and have argued on its behalf.
Q: How about the technology?
A: The technology, I think, is somewhat inevitable. Oracle has made acquisitions in the area recently, and as a variety of folks here at the show pointed out yesterday, Informatica has many of the necessary pieces (making it a potentially more attractive acquisition target, in light of this announcement).
I don’t mean to imply that the information server is likely to displace the database or anything similarly radical – the standalone database will be around for a long time to come – but I think you’re already seeing many of the larger data management firms augment their stores with a variety of value add technologies.
Q: Can you sum up your reactions?
A: My initial affection for the term information server has not dimmed at all. I’m a believer in the idea, and look forward to seeing the product set – from IBM and from competitors – evolve. In the meantime, I think IBM has put together an interesting offering for higher end customers, and I’d recommend they give it a look. For lower end customers, I think the wait will be significant.