An InterSystems Technology Summit

I have a great love for interesting, data-focussed, technology tool vendors that have a respectable and intensely loyal customer-base – and a very low profile in the press. Usually, they are well-established (often with, dare I say it, an old-fashioned look and feel), and privately held (or, sometimes, a small part of a larger publicly-quoted company). They usually succeed (or survive) by solving real problems other, better-known, tools can’t solve, but are hard to market generally because general developers without a failed project on their hands object to their idiosyncratic look-and-feel and give up on them (and probably disparage them on the Web, which doesn’t help) before discovering what they can do. Vendors like Progress, Uniface, Rocket Software and InterSystems come to my mind – but you should check up on the quality of their customer lists sometime.

I’ve just been at the InterSystems Technology Summit 2015 – listening to a new executive rejoicing in the fact that he was at the conference talking to customers and keeping them happy rather than doing unnatural things with buying departments in order to make his Q2 sales quota – and keep the stock exchange happy. As he would have been if he’d still been with his previous, publicly-quoted, employer. Companies that don’t have to keep stock market analysts happy can concentrate on customers.

And InterSystems had some good stories to tell its customers this year, starting with a new Eclipse-based IDE, called Atelier (betas available soon), which will eventually be an alternative to its current Studio tool. This should make marketing to new developers that don’t know InterSystems already a lot easier (compare Uniface 10, although its modernised IDE isn’t Eclipse-based).

InterSystems chose Eclipse for its new IDE largely because it is cross-vendor and multiplatform, and because it gives it access to a larger ecosystem. InterSystems, for example, currently stores its source-code as XML in its database, which has consequences (what you see isn’t always what you get, cosmetically, when you retrieve it) and only provides basic SCM (Source Code Management) functionality; but Atelier can use any SCM repository supported by Eclipse plug-ins – Git, Perforce and so on. And InterSystems doesn’t see writing SCM tools as its core business.

Certainly, Bill McCormick, Director of Product Management (responsible for managing the existing Studio programming interface) showed palpable enthusiasm for the responsiveness of the alpha version of Atelier he was demonstrating (unlike Studio, it is client-based rather than server-based), and the way it lets him format nice readable code – and get it back formatted the same way from the repository. In addition, McCormick apparently sees Atelier as a chance to re-factor lots of features of Caché programming that “just growed” and could be done better. I’d love to see the capabilities of InterSystems Caché database more widely exploited, and I think that the availability of Atelier should remove one obvious barrier to wider adoption.

InterSystems Caché itself is now going to be marketed as part of a bigger data platform – which can help you innovate; using all of your data, not just the data that fits into a relational table. It also sits within the InterSystems Ensemble and HealthShare platforms that I may talk about another time. So, for example, a company called PCS uses it to build a content management system for print and online journalism (an innovative way of making print journalism relevant in a digital world) and it uses InterSystems iKnow, one of its platform capabilities, for natural-language analysis that automatically generates concept metadata from the text content (that can be stored in Caché).

The Caché Platform, with an an open architecture, can be used to build pipelines that combine its storage and other platform facilities with those from third parties, to deliver business outcomes. So, when the PCS story made me think about generating metadata automatically for photographs, the InterSystems response was that iKnow could do that now, using free-form text descriptions of a photo – but that if I could find a specialist image analysis program by a 3rd party, I could easily feed this into iKnow too. Whether I could interface iKnow to, say, Adobe Lightroom, depends on how open Lightroom is. One important standard here, for which InterSystems is developing support, is Oasis UIMA (Unstructured Information Management Architecture), which (amongst other things) will allow its natural language processing to be in a pipeline with other tools.

Caché has amply demonstrated its ability to scale up for volumes well beyond anything it needs to deal with today. Check out this white-paper, for example. Carl Dvorak is President of Epic Systems Corporation, a privately held healthcare software company that claims to hold the medical records of just over half of US patients, and he says that: “the scalability and performance improvements of Caché version 2015.1 are terrific. Almost doubling the scalability, this version provides a key strategic advantage for our user organizations who are pursuing large-scale medical informatics programs as well as aggressive growth strategies in preparation for the volume-to-value transformation in healthcare”.

Nevertheless, the story is about a lot more than just storage capacity and processing speed. iKnow now speaks Russian and Ukrainian, so the Caché store used by the Ukrainian Justice System (from within an InterSystems Ensemble-based application suite) is used to support the sort of legal transparency needed if it is to join in a European community; but as well as iKnow for natural language processing, the Platform also supports DeepSee, for rich analytics and can take advantage of InterSystems’ experience with medical device connectivity – which should be reusable for the IoT.

The InterSystems future product roadmap makes interesting reading. It is aiming at an entirely open architecture, and “multi-model” database support (relational, graph, document etc.), with SQL as a lingua franca and NewSQL (i.e., modern and competitive) performance levels. Caché won’t go away; and underneath the covers of a new product, users of “legacy” Caché systems will still be supported. Intersystems platforms can do this because they have a strong segregation between the logical data model and the underlying physical storage. In the same way that they can efficiently implement an SQL view on top of the underlying, physical, sparse array, without compromising performance, it should be able to overlay a graph database view, say, as well. And, of course, Caché has been, technically a NO-SQL (in the sense of “Not Only SQL) database for ever.

If InterSystems really delivers on its new open model (and that may be a cultural thing as much as a technical one; since one of the risks for companies like those mentioned at the start of this article is that they can get a bit self-centred) – and since there’s no reason why it shouldn’t deliver – we could see InterSystems vying with the Big Three database systems for visibility in the future! InterSystems is also thinking of using open source to complement its platform offerings, not for the database and core offerings but for some of the associated services, although there could be cultural issues here too (the Open Source Software community can be quite unforgiving). Atelier, for example, will have an open API, so that anyone who doesn’t like InterSystems’ IDE choices could simply write their own.

InterSystems is now in a good place, I think,  to exploit the current state-of-the market: new databases and database models, in addition to the relational model, are becoming fashionable at last; as analytics and IoT trends place increasing strains on pure relational products.