I was recently approached by a healthcare company for advice on suitable database solutions capable of executing its new initiative. The company was primarily an Oracle shop so naturally, they began by shopping for possible Oracle solutions.
The CTO relayed his conversation with the Oracle sales rep, who at first recommended an Oracle solution that, expensive as it may have been, ultimately aligned with the company’s existing technology and experience. Unfortunately this didn’t match their budget and so predictably, the Oracle sales rep whipped out a MySQL-based solution as an alternative.
Having worked as an Oracle DBA throughout the dot-com years, I know the technology well. I also know the cultural differences between enterprises that choose Oracle solutions and those that choose open-source ones.
This encounter with the healthcare firm struck me as a classic conundrum for today’s companies who are under pressure to meet business targets under a tight budget, and in a very short time.
Can an open-source solution like MySQL be the answer to such huge demands?
The Oracle sales rep will likely nod excitedly and say no sweat. But as a consultant I could only manage an equivocal yes.
In my experience with startup after startup I’ve seen plenty of different MySQL installations but I’d never heard of one with the technology stack he described. What’s more I’d never heard of these solutions described with the Oracle Corp titles.
On one hand I wanted to discuss the merits of the solution he was keen to implement, while on the other, I was expressing concern over possible directions and paths we might take.
An Oracle cluster is not a MySQL cluster
The solution Oracle suggested was a MySQL Cluster. The term cluster unfortunately means different things to different people. Such loose usage of the word dilutes its meaning. In particular a lot of Oracle technologists expect that this solution might be similar to Oracle’s Real Application Cluster technology. It’s not. There are a lot of limitations, and frankly it’s really just a different beast.
The list also included various management dashboards which Oracle likes to push, but which I rarely see in my consulting assignments. What’s more I heard nothing about replication integrity considering that replication problems are an ongoing concern for real-world MySQL installations due to the particular technology used under the hood. There are reliable solutions to this problem but none yet available from Oracle. In fact, this is a big problem but one that may be completely off the sales guys’ radar.
Don’t let sales frame your architecture
Honestly, I don’t have a particularly large axe to grind with the sales guys. They have a job to do, and providing solutions which bring revenue to their firm and commissions for themselves is what puts food on their tables. Each party is motivated in different ways. But as a company shopping for solutions, this should be kept clearly in mind when starting down that road.
Beware prescribed architectural frameworks that appear too easy because they almost always don’t do what they say on the tin. Unfortunately sales folks don’t have experiencing designing architectures in the real world, so they can’t really know how the technologies work beyond the data sheet with feature bullet points.
As we all know in the technology space, all software come with bugs and real-world experience does not match the feature lists in the brochures. In law they have de jure and de facto. The former describes what is written and the latter, what’s practiced. For technology solutions, its never just adding water for something to work.
Do your homework
Before you embark on a new trip through the open source technology jungle, do some due diligence. Read up on real-world solutions, and how other large firms are using the technology. What configurations are they having success with? Which are causing trouble for a lot of people.
One of the great advantages of open-source are the very vibrant communities, forums and discussion groups where people are glad to share their experiences and offer advice.
Allow sufficient time to test and
bring your team up to speed
This is very important one. Shifting from an enterprise that relies primarily on Oracle for it’s relational database solution over to one that relies on open source technologies is a very big step indeed. Open-source technologies tend to be much more do-it-yourself and roll your own. Oracle solutions tend much more toward predefined paths and solutions and prescriptions for customers.
There are merits to each of these paths, with attendant pros and cons. But they are decidedly different. It’s likely that your team will also require time to get up to speed, not just with the particular software components, but with the new process by which things happen in the open-source space. Allow sufficient time for this shift to take place, lest you create more problems than solutions.