Yesterday IBM once again showed how serious it is about taking on the likes of Amazon Web Services, Microsoft Azure, Oracle Cloud, Salesforce 1, and SAP Cloud as a general purpose, global, cloud business platform. Between the unveiling of the BlueMix beta for cloud application service composition (aka application development) and the announcement of the pending acquisition of Cloudant, a Cloud database service, IBM has added, using its own R&D plus M&A, fresh ingredients to its PaaS mix.
Of the two, BlueMix seems the more daring, the result of IBM rethinking what developers must do when creating cloud apps. Leaning to some degree on open source Cloud Foundry, and pushing the idea of visual development environments in an entirely new direction, BlueMix enables developers to stir together enticing application compositions through:
* Visual discovery and sharing of app and middleware (e.g. database) services with,
* programming language and tool choice, plus,
* straightforward cloud deployment awareness, provisioning, and flexibility.
IBM ostensibly targets “line of business developers” with BlueMix, and certainly the compositional, almost social, nature of BlueMix will prove appealing to enterprise business app developers. Longer-term, however, it seems conceivable that a full range of app developers will use BlueMix as a visually compelling, cloud-oriented kind of GitHub. Time will tell whether BlueMix will harden into a true enterprise-grade development environment, but it is doubtful IBM would have gone this far without expecting BlueMix to succeed commercially. “Succeed commercially” does not mean, however, selling BlueMix subscriptions and licenses, but rather means compelling customers towards IBM SoftLayer and related services.
On the surface a curious choice by IBM for its NoSQL Cloud database, before yesterday Cloudant was just one of a big inventory of tier-2 databases fighting for market space. From the Boston area alone, which has largely hosted the rebirth in database technology, with many vendors following the commercial/open source co-dependent model, Cloudant was one of the least known. Cloudant leverages open source Apache CouchDB, a schemaless JSON store, and specifically the BigCouch variant which Cloudant merged into CouchDB. IBM might have also considered, from Beantown alone, entries like Basho (based on open source Riak), EnterpriseDB (PostgreSQL), NuoDB (Neo4j), Paradigm4 (SciDB), Sqrrl (Accumulo), Tesora (OpenStack Trove), and VoltDB (PostgreSQL roots). In addition, IBM has actively partnered with MongoDB, the acknowledged 800 pound gorilla of the NoSQL database movement, and likely also considered the strongest of the Silicon Valley NoSQL players including Couchbase, DataStax, and MarkLogic.
There seemed little doubt, however, that IBM needed a cloud NoSQL database of its own. DB2 and Informix could not answer all cloud callings. Oracle had begun to push its own Oracle NoSQL database more actively in its channel, adding the NoSQL play to its new Oracle database 12c clearly targeting cloud, plus Times Ten, MySQL, open source Berkeley DB, and its database appliances. IBM needed some response to that line-up, and NoSQL seemed the obvious starting place. There was no reason for IBM to develop a NoSQL database from scratch, given all the options in the market. Why then Cloudant?
With the possible exception of Basho, Cloudant was the only NoSQL entry born and raised with the express intention of offering a multitenant, elastic, scale out, global cloud database service. As Cloudant CEO Derek Schoettle stated, “Scalability is table stakes” in today’s cloud environments. Thus, Cloudant also includes the geo load balancing and replication required for global Web apps.
Offered as a true DBaaS, Cloudant already operates in 34 data centers around the world. Most importantly, Cloudant is already widely deployed on SoftLayer. Customers also deploy Cloudant on dedicated cloud infrastructure, enabling the company to deliver its managed service on a larger scale. This nicely fits IBM’s customer-friendly approach of “Have cloud your way,” be that public cloud, private, hybrid, or managed service. Note that Cloudant was so cloud-centric it didn’t even offer on-premise style licensing.
It will be interesting to see whether IBM continues offering Cloudant on Microsoft Azure and Heroku, two other PaaS clouds where Cloudant has gained a foothold. This represents a tricky choice for IBM – do you want to keep Cloudant strictly on your own cloud as your native NoSQL service, or popularize Cloudant on other clouds? No opinion here; glad the choice is IBM’s.
In terms of apps, Cloudant only supports eventual consistency and thus is not the best choice for transactional apps. Despite some positioning of Cloudant for big data analytics, it more optimally fits Web scale apps, search, mobile apps, gaming, and document/media oriented solutions. Does Cloudant’s lack of true transactional support portend another ACID NoSQL acquisition in IBM’s future? The rest of the Beantown-based and Silicon Valley NoSQL providers that list “ACID” in their features may want to ensure their Smartphone’s battery remains fully charged.
Summary: Safe Ingredients, After All
The takeaway for IBM customers and cloud prospects regarding Cloudant, after further examination, seems obvious: It works. IBM doesn’t have to pour a single cent of its promised $1b allocation for cloud software into making Cloudant work on SoftLayer. Let the marketing and sales begin. IBM, however, may want to back-fit some of its development tools to better support Cloudant, which brings us directly back to BlueMix.
In the IT industry, the count of software developers actively working on a platform is the best non-monetary metric of platform success. Windows 3.1, Java, iOS, and Android all “won” because they wooed the most developers. A similar battle now brews between clouds. From that perspective, BlueMix may end up being the most strategic investment, tied with SoftLayer, IBM has yet made in its attempt to gain the most business cloud market share.