We now return you to your regularly scheduled program.
Amazon released the latest offering in their AWS suite last month, SimpleDB. I tagged this and subsequent analysis posts for later consumption, and I’m just now getting back to it; a post from Web Worker Daily yesterday on LongJump and their new Database-as-a-Service offering jolted my memory.
I pinged Rich to get some thoughts before starting, and he pointed me to CouchDB, an open source project very similar to SimpleDB that’s been around longer. He also clued me into IBM’s recent hiring of the lead developer for CouchDB, Damien Katz, whose role will not change much at IBM, i.e. the CouchDB code will be donated to the Apache Foundation and remain open source, and IBM is serious about open source and service offerings. Damien addresses the FAQ here.
As I read through the coverage (some of the highlights), including an interesting idea by Bex to layer Oracle Coherence on top of the S3/SimpleDB combo, it became clear that my two technical cents would be woefully inadequate. Instead, a few other things bubbled up:
- “As-a-service” offerings were the Holy Grail back in the late Bubble, and we’re finally seeing why.
- Amazon has a gold mine, and watching this play out will be fascinating.
- Starting an Interwebs company is too easy, contributing to Lazyweb.
What it means to startups
Craig Cmehil, Ethan Jewett and I had a conversation recently about how-to Enterprise 2.0; in the comments on Craig’s post, the issue of scant hardware resources was discussed. When we built Mix, obtaining hardware was among the toughest tasks. Plus, once we had the hardware, Rich had to take a crash course in Oracle database installation. We inquired around to see if anyone ran a database farm we could use, but no dice.
These issues are not new to startups. Back up to the fallout after the Bubble burst, and you could see what a lot of dot-coms spent on when they auctioned off their Cisco network gear and Sun servers. Obviously, you didn’t see the software licenses, but those were contributing factors too. Hosted services were gaining widespread momentum in 2000, but a lot of startup went the self-maintenance route.
I worked at a data center in 2000 in product development, and the plan was to recoup the sunk costs of raised floor, rack space and power/ping/pipe by selling managed services, i.e. applications, database, whatever, to customers. When the Bubble burst and our customers stared going out of business, that plan was moot.
Fast forward to the present, and it’s a reality through open source products, hosted application suites and services from Amazon (and others) like SimpleDB, EC2 (Elastic Compute Cloud) and S3 (Simple Storage Service). The vision of 99-00 is finally being realized, and startups no longer need to raise eight figure rounds to cover infrastructure and operations growth.
What it means to Amazon
Now, Amazon has a one-stop shop for startups. Companies like Twitter (which uses S3) can focus on the business, not that Twitter has a business plan, but in theory, this is what they could do. What’s weird about this to me is how did the web services vision coalesce at Amazon?
I would love to have been a fly on the wall in the meeting when this was pitched to Jeff Bezos, or maybe it was his idea. Either way to see the looks on the faces in the room would have been priceless. After all, web services seems like a happy accident, based on Amazon’s primary business, but I guess once you build tens of thousands of feet of raised floor space and line it with thousands of rows of racks and put gear in all the racks, only to find out you’ve got some excess p/p/p what else would you do with it?
While SimpleDB is not technically a database service, it seems pretty close. Topper started a question thread on Mix called “How do we attract a younger generation to the Oracle World?“. I think one way is to offer up giant database farms for cheap. This simultaneously exposes younger people to Oracle, by lowering the cost and raising the utility, and strikes at MySQL. Insert disclaimer here: I don’t speak for Oracle. These are my opinions.
What it means to development
I blogged about the lazyweb a few months ago. Zed’s rant against the Rails community reminded me of the lazyweb, i.e. if Rails is so easy that it removes some of the barriers to entry, like education, training, experience, can that be a good thing? Doesn’t this lead to bad code and bad applications?
Now, I’m wondering the same thing about everything “as-a-service”. It’s a good thing to remove the sunk costs associated with starting a web-based enterprise. Operations and infrastructure are best handled by professionals who have paid the big bucks for big iron and state-of-the-art security. But SimpleDB gets into design, so is it good thing that Amazon has already architected the guts of your data store? Hmm.
I’m not sure, and I guess we’ll see as it matures. One thing is for sure, the wins and fails will play out for all to see and analyze.