1 CS : Technology Trends Ion Stoica ( September 12, 2011
2 “Skate where the puck's going, not where it's been” – Walter Gretzky “Skate where the puck's going, not where it's been” – Walter Gretzky
Processors MIMD (Multi-Core Processors) – linear increase: two additional cores every two years SIMD (GPUs) – exponential increase: width double every four years 3
SSDs Performance: Reads: 25us latency Write: 200us latency Erase: 1,5 ms Steady state, when SSD full One erase every 64 or 128 reads (depending on page size) Lifetime: 100,000-1 million writes per page 4 Rule of thumb: writes 10x more expensive than reads, and erases 10x more expensive than writes Rule of thumb: writes 10x more expensive than reads, and erases 10x more expensive than writes
Storage Performance & Price Bwdth (sequential R/W) Cost/GBSize HHD MB/s$ /GB2-4 TB SSD MB/s (SATA) 1.5 GB/s (PCI) $2-4/GB200GB-1TB DRAM10-16 GB/s$12-13/GB64GB-256GB 5 Bwdth: SSD up to x10 than HDD, DRAM > x10 than SSD Price: HDD x20 less than SSD, SSD x5 less than DRAM Bwdth: SSD up to x10 than HDD, DRAM > x10 than SSD Price: HDD x20 less than SSD, SSD x5 less than DRAM 1
Storage Price Trends RAMs: x2 every ~20 month ( : x75 decrease : x63 decrease Disks: x2 decrease every ~2 years SSDs prices dropped faster than disk prices for last 5 year ( trends-of-hdds-vs-ssds/) ( trends-of-hdds-vs-ssds/ But decrease slightly less over last year 6 Storage price halves every ~2 years
Hard Drives (25 years ago) IBM Personal Computer/AT (1986) 30 MB hard disk - $ ms seek time MB/s (est.) sec to scan entire disk
Memory (today) 96 GB RAM - $ (ECC RAM) Memory bus speed: GB/s sec to scan entire memory!
Working Set: RAM Doubling Software (1995) 9
Working Set (Today) When was the last time your experience trashing on your laptop? Memory growing faster than application’s needs – conjecture 10 Today’s memory, yesterday’s disk!
Working Set – Datacenters % of jobs whose full inputs fit in memory (~1 week) 11 Memory (GB) Facebook (% jobs) Microsoft (% jobs) Yahoo! (% jobs) Nearly all jobs’ inputs fitting in main memory in near future? Nearly all jobs’ inputs fitting in main memory in near future? (Ganesh Ananthanarayanan)
(Random) Thoughts Today’s disks, yesterday’s tapes [John Ousterhout] Today’s memory, yesterday’s disk? Or should be today’s SSDs, yesterday’s disks? SSDs not great for caches (due limited writes) Perfect for archival though and GFS-like filer systems ;-) In-memory computation not enough for interactive workloads Parallelism only way out if need to touch a lot of data 12
(Random) Thoughts (cont’d) Today’s servers in Hadoop clusters: disks Up to 1GB/s bwdth How to take advantage of this? GPU use will only increase: faster increase in processing power than CPUs Need better support for virtualization What to do about memory bwdth? For data intensive apps, locality will continue to be critical 13
Predictions?? Memory the new disk Working sets of more and more apps will fit in memory SSDs will become the new tape (archival) GPUs: main driver for increasing processing power Will be integrated in the main processor 14