2010: The year of the microslice server
perspectives.mvdirona.com
perspectives.mvdirona.com
Currently, your typical dual-socket, quad/six core CPU, is quite top heavy: LOTS of CPU performance, only a satisfactory amount of RAM bandwidth (and in certain cases not limited by bandwidth but by the DRAM command rate), along with piddling disk and network performance.
You could create a dual or quad-core ARM Cortex A-9 chip with a dual-channel memory controller (more memory bandwidth/ops per degree of CPU performance), dual or single gigE (more relative bandwidth per node), and finish it with a single SSD (lots of IOPs). Since these nodes are fairly affordable (probably $500) and relatively low power (probably 10-15 watts, you're left with nodes that 1) offer 1/3rd the performance but are 1/5th the price and use 1/20th the power 2) can be scaled cheaply and predictability, particularly as it pertains to laying out your datacenter power and cooling infrastructure.
Also, for people who don't have complete control of their data center (they're renting rack space), it eliminates a significant amount of headache finagling your datacenter to provide enough amperage to your rack, and eliminates surprise deficiencies in their cooling infrastructure.... lessons I've learned the hard way. Micro-servers let you scale at a much more granular level so that you aren't rolling out several kilowatts of servers at a time, and let you avoid carrying excess capacity as you wait to grow-into it (only to scramble to scale after you're full-up again)
It can be hard to get over the sense that always compressing and decompressing everything that goes to disk is wasteful. But available CPU cycles left unused are marginally free and can't be stored us for use after the moment passes. They should be "wasted" to achieve even a slight improvement in overall throughput.
Overall I wonder if maybe a new OS for the cloud is called for? It seems inefficient to have seperate VMs running full OS for every tiny application. Maybe in the future not only storage will be a service (like S3), but also CPUs and RAM could be plugged together at will? Like there wouldn't be lots of small or big server instances, there would be farms of CPUs, farms of RAM, farms of storage, that could be combined at will. Maybe networking would be too much of a bottleneck, though :-/
tl;dr IO is the bottle-neck, not the CPU. So slow CPUs on lots of nodes gives you more cost-effective (in terms of power costs) key-value lookups.
Of course, this makes layer-7 load balancing a huge problem, which is my own little project in this world. More on that later :)
Dual core atom processor 2 GB Ram (expandable to 4gb) Solid state disk for OS Mini itx form factor
Supermicro 1U 230 barebones chassis is $200 http://www.newegg.com/Product/Product.aspx?Item=N82E16816101...
$50 for ram and <$100 on storage for a HDD or $200 for a small SSD. Bulk prices would take that down by probably 10-20% today so you could probably do this for $300 today. I believe Dell is selling systems like this as private integrations just not retail, I'd imagine their costs are substantially lower.
Bump the CPU to a 330 or whatever, bump the ram sure but you're still talking $300 when a normal 1U HP server is <$1k list and is vastly more bang for buck. The cooling and power density arguments aren't going to be very persuasive (IMO) unless the price is there as well, hence my hope for very low per-unit costs, even if there are chassis/bladecenter components as part of the package.
It sounds like they're finally willing to open this up and allow a decent amount of ram (8GB and beyond) on low-power, tiny motherboards.