On Wed, 2007-10-03 at 14:10 +0800, Orlando Andico wrote: > 1) if you use a DB, the DB becomes a Single Point of > Failure, and a Single Point of Scaling bottleneck > > 2) NFS! locking issues galore! horrific file lock > timeouts! I can't imagine this scaling up well
Yeah, it's useful to know what the limitations are. As with all discussions like this one, there are never any absolute answers, everything depends on the context. and the context is different for different people unless everyone agrees on a theoretical context for purposes of discussion. I've not worked with any system that needs to keep up with even 60 requests per second, so I don't yet need anything that will scale to that level. DB and NFS are good enough for everything I've needed to do to date. When I need more reliability and performance than those can provide, well, I'll look into the options hopefully six months before I need it :-). With respect to the DB being a single point of failure and bottleneck. Normally it's already that anyway. Either you accept it (because it's not a problem yet) or you workaround (with some sort of DB replication system). I certainly haven't yet seen anything like your example previously where the transaction will rollback AND RETRY if the apache process dies or the database goes south. rollback isn't hard. retry is definitely interesting though. > You don't hear a big deal about it because the people who > need the performance (Yahoo, Google) even though they use > FOSS, they keep their implementations close to their chests. Yeah. Although possibly bits and pieces of come out once in a while (e.g., the projects at skype) http://joseph.randomnetworks.com/archives/2007/03/15/postgresql-projects-at-skype/ quite a lot comes out of google too. But not packaged together as a coherent whole. > If you use NFS or DB for scaling, your application is > probably small enough that you don't see the scalability problems. > > I think that's even more dishonest than people like Intel or > Microsoft. At least Intel and Microsoft are honest about their > software. I'm not clear on who you mean is being dishonest there or what that someone is being dishonest about. > Agreed. I used to roll creaky solutions right and left. > And at the time I thought I was being excessively clever. :-). And having some amount of fun too, probably. It's good to be doing that when young. As you gain, ahh, experience, things change. I don't think you have a baby yet, Orly, things change even more then :-). > I guess if you don't have that huge a revenue, you can tolerate > outages and less-than-enterprise class software. It's a tradeoff. Everything is a tradeoff. The trick is *understanding* the tradeoffs. If you buy the tradeoff without understanding it, then you may find that you've bought the wrong thing and that you should have paid money for better. On the other hand, if you understand the factors being traded off and you can accept that level of risk then at least you had your eyes open and accepted the risk consciously (and got your boss to understand and sign off on the risks). People running an open source skunkworks without the knowledge or approval of their bosses are on their own if things break and they can't bring things back up before the bosses notice :-). tiger -- Gerald Timothy Quimpo [EMAIL PROTECTED] Business Systems Development, KFC/Mr Donut/Ramcar I don't know half of you half as well as I should like; and I like less than half of you half as well as you deserve. -- J.R.R. Tolkien _________________________________________________ Philippine Linux Users' Group (PLUG) Mailing List [email protected] (#PLUG @ irc.free.net.ph) Read the Guidelines: http://linux.org.ph/lists Searchable Archives: http://archives.free.net.ph
