|
Originally Posted by Clarity
One thing that I think we probably should do, is figure out *exactly* what a new system is going to cost. I've seen a lot of differing viewpoints on this, ranging from $3k (what the current machine cost 3-4 years ago) to $12k (what our target machine would cost from Dell).
I am, as you each know, *not* the tech type. So let me lay out the issues, and maybe someone can get a group of tech-minded people together in a little server summit of sorts, to decide what makes the most sense. From there we have a specific cost, we can post that during the fundraiser as a real world target, and even offer some sort of progress meter.
Where we are hosted is the highest quality bandwidth, and it is free. The trade off there is we have zero support. If something breaks and requires someone to actually do something to the machine itself, there's really no telling whether that action would come in minutes, hours, or days. Yeah, scary, I know.
So what we need first and foremost is RELIABILITY. Really no question there. We need a stable well-built machine that doesn't have a lot of guesswork involved. A box that will come online, and run for years without a reboot (as this current one, miraculously) has. The current machine will stay online as a backup of sorts, so once we have a second machine, the danger of a machine needing attention will be greatly reduced, but it's still our highest priority.
Second is RAM. I've seen quite a bit of conflicting viewpoints on this. I saw someone saying they don't understand why 2MB isn't enough, and they work for an ISP. I saw someone else saying 4MB should be enough.
Here, I have to go with the thinking of two groups. JelSoft (the people who made vBulletin) and our sats engine programmer (damn do I need to touch base with him, he probably thinks me (and the project) dead). They've both recommended 8MB. Both said that we would be fine with 4MB, but would need to add to that in the future based on site growth and how heavy traffic to the stats engine becomes.
I look at this a bit like highway projects. You have a 4 lane road, 2 in both directions. The city decides to add one more land going in each direction, because 4 lanes is far less capacity than the city currently needs. So over a 2 year period, they complete the construction and it's now a 6-lane road. But growth in the area has, in that time, exceeded that need, and while 6 lanes would have solved the problem 2 years ago, now it really needs 8. So they embark on yet another construction project to bring it up to speed.
Maybe that doesn't make sense. Lol. Basically, 8MB is saying "we're good now, and for years to come." 4MB is saying "we're good now". But the issue is that costs literally explode as you add RAM. So I don't know.
Processor is one of the least important areas. Slap a single P4 3.0+GHz chip in there, or a pair, and we're perfectly fine.
Hard drives - the current machine is running on a trio of 18GB 15k rpm drives in a RAID-5 configuration. We could easily go to a pair of bigger and slight slower drives for cost purposes. Say, two 100GB 10k rpm drives in a RAID-0 (or is it RAID-1?) configuration where one is a live/simultaneous backup of the other. So if one fails, the other stands in immediately. Cheaper, and less complicated, but now we are *way* out of my area of technical competance.
Besides that, it needs to have a high speed network card or two. It needs to be rack mountable. I'm not sure we have any real needs beyond that point.
|
03-14-2005, 01:24 PM
Linear Mode
