Scalability by adding boxes
December 7th, 2007
Not that I know too much about payment processing at PayPal, but each time I read a story like this that says “we’re replacing System i/p/z with System x boxes and it’s so much cheaper” I always wonder two things.
1) As a payments company do you really want to “follow the Linux kernel development process” just to process a few transactions per second? Is it really a prerequisite to have people that know the kernel? A moment ago we were still talking about a payment system, not a device driver.
2) Does it sound sensible to run thousands of servers and maintain the staff and software to actually do it? – I have to say that I always found it interesting to see how much workload large systems can process and how few people you need to do it.
December 7th, 2007 at 10:00am CET
hmm, you left some crucial factors in the dark i guess.
*) employee costs
this factor is quite high for any company, at least if we’re talking about highly professional skilled employees. however you have a lot of options to spread this costs in most countries.
*) virtualization
i’ve had a paper in my hands, some days ago. a quite big financial institute that reduced their serverfarms through an average factor of 24. you can scale much more dynamically with such an approach.
*) linux kernel myth
do you really think, most companies do care about that? i guess, they don’t. they simply use operating systems that cover this risk to a high grade through bundling appropriate hardware. meaning, like a lot of that companies use centos because they can order their servers from dell which have full driver support.
December 7th, 2007 at 11:22am CET
@Wolfgang: Great points. – The reason I mentioned the linux kernel btw was because this was mentioned in the article. – I “hope” that most customers, even if they use linux, won’t care about this. – Not that anyone cares about POSIX these days, but most likely because everyone uses Java :-)