Elastra – somewhat more sensible Amazon-based DBMS option
Elastra is a startup offering MySQL and PostgreSQL SaaS instances in the Amazon S3/EC2 cloud. On their board is John Hummer, which I generally regard as a good thing, although it’s hardly a guarantee of success.* High Scalability raises some doubts about Elastra’s pricing, but I think that may be missing the point.
*John’s track record includes Powersoft, Octel, and Employease — but also Pets.com, Napster, a company that gave me a ridiculous briefing, and a .524% career NBA free-throw percentage.
In a post earlier tonight about Amazon’s new SimpleDB, I suggested that SimpleDB’s main use might be as the database engine behind other S3/EC2 apps. In such scenarios, you don’t necessarily need a big or high-volume database; you just need one that works occasionally to accept orders and so on. On the other hand, you probably do want transaction integrity and other platform features that have been standard since at least the Reagan Administration. And in cases like that, Elastra’s service might not be too costly.
Of course, the same reasoning suggests that there might be trouble finding a hugely rewarding revenue model for it, but that’s not a big user concern. If the worst thing that happens to you is that your SaaS provider fizzles and you eventually have to find another way to host MySQL or PostgreSQL online, you should be fine …
Comments
2 Responses to “Elastra – somewhat more sensible Amazon-based DBMS option”
Leave a Reply
Hi Curt,
First, thanks for the nod.
I’m very excited to hear about SimpleDB and from a marketing fluff perspective it helps raise awareness of putting one’s data in the “cloud.” Early adopters will no doubt be excited by this low cost entry as well.
In the online database space, I do see SimpleDB as another option for many developers. For comparison, for better or worse, SimpleDB resembles Berkeley DB so the decision to use either or should be familiar. Once you go that route, your somewhat committed.
SimpleDB, while not BDB exactly, has similar characteristics except for one, very important feature: Eventual Consistency. I personally think that this concept of eventual consistency will require some wrestling on part of the application developer. If someone were able to write a SQL interface on top of SimpleDB to leverage those existing tools like ActiveRecord, Hibernate, JPA, etc. the biggest challenge will be rectifying the eventual consistency constraint. No longer can you assume that a write followed by a read will yield what was written.
Does this mean that SimpleDB is not suitable for any problem? Of course not. Internally at Elastra, we currently have a project that can accommodate eventual consistency and the notion of using SimpleDB is extremely compelling (my beta key is coming any day now ;).
At Elastra, we think that the database is a valuable pillar of the software stack Data warehousing and decision support software must rely on the semantics of relational databases. Transactions are a good thing. Offering a cost effective service that provides a hosted relational database that can be dynamically scaled is attractive to many people, especially those without DBAs.
We also hear a demand for other services common in the enterprise to offer in a similar on-demand model. Provisioning a database requires simply asking for a database engine (MySQL and PostgreSQL) and pulling the trigger. Pay for only what you need. You need an Apache farm or a ESB? The same model can be applied.
Our pipeline for enterprise applications is exciting and I’m looking forward to the offering. In almost all cases, the database is a somewhat more sensible place to start for these services.
-Noah Campbell
VP of Engineering
Elastra Corporation
Hi Noah,
Thanks for stopping by. If you’d like to tell me about Elastra in more detail, please go ahead and have your “people” 🙂 schedule a call.
In particular, since lots of high-end hosts support DBMS — duh — I’d be curious to learn more about your value-add.
Thanks,
CAM