DigitalOcean SFO1 vs SFO2
DigitalOcean SFO1 vs SFO2
Yesterday DigitalOcean announced the availability of their SFO2 datacenter. Curious to see what was different, I did a quick investigation. This is by no means exhaustive, and I’m a more elaborate application specific benchmark would be more interesting, but this is a quick overview.
Network
Quick check from my apartment in San Francisco on Comcast (and via WiFi, I know…).
SFO1
$ tracepath 45.55.15.x
1?: [LOCALHOST] pmtu 1500
1: router 0.810ms
2: 172.16.12.1 1.446ms
3: 96.120.89.209 12.058ms
4: te-0-7-0-4-sur04.sfgeary.ca.sfba.comcast.net 10.930ms
5: be-313-ar01.hayward.ca.sfba.comcast.net 12.025ms
6: lag-14.ear2.SanJose1.Level3.net 14.050ms
7: ae-1-6.bar2.SanFrancisco1.Level3.net 15.104ms asymm 9
8: DIGITAL-OCE.bar2.SanFrancisco1.Level3.net 19.098ms asymm 11
9: no reply
10: 45.55.15.x 15.841ms reached
SFO2
$ tracepath 138.68.15.x
1?: [LOCALHOST] pmtu 1500
1: router 1.009ms
2: 172.16.12.1 1.294ms
3: 96.120.89.209 13.694ms
4: te-0-7-0-4-sur04.sfgeary.ca.sfba.comcast.net 11.516ms
5: be-313-ar01.hayward.ca.sfba.comcast.net 12.551ms
6: lag-14.ear2.SanJose1.Level3.net 14.765ms
7: DIGITAL-OCE.ear2.SanJose1.Level3.net 12.979ms
8: no reply
9: 138.68.15.x 14.680ms reached
Resume: pmtu 1500 hops 9 back 9
Nothing really interesting to see here.
Processors
How about the processors? Did they move to a new generation?
SFO1
root@ubuntu-512mb-sfo1-01:~# cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 45
model name : Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz
stepping : 7
microcode : 0x1
cpu MHz : 2299.998
cache size : 15360 KB
physical id : 0
siblings : 1
core id : 0
cpu cores : 1
apicid : 0
initial apicid : 0
fpu : yes
fpu_exception : yes
cpuid level : 13
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon rep_good nopl eagerfpu pni pclmulqdq vmx ssse3 cx16 pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx hypervisor lahf_lm vnmi ept tsc_adjust xsaveopt arat
bugs :
bogomips : 4599.99
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management:
SFO2
root@ubuntu-512mb-sfo2-01:~# cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 63
model name : Intel(R) Xeon(R) CPU E5-2650L v3 @ 1.80GHz
stepping : 2
microcode : 0x1
cpu MHz : 1799.998
cache size : 30720 KB
physical id : 0
siblings : 1
core id : 0
cpu cores : 1
apicid : 0
initial apicid : 0
fpu : yes
fpu_exception : yes
cpuid level : 13
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon rep_good nopl eagerfpu pni pclmulqdq vmx ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm vnmi ept fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid xsaveopt arat
bugs :
bogomips : 3599.99
clflush size : 64
cache_alignment : 64
address sizes : 40 bits physical, 48 bits virtual
power management:
The SFO2 datacenter might actually preform slower then the SFO1 datacenter because the chips (after updating the Intel ARK link) are actually clocked slower and are of the same generation. DigitalOcean traded per core performance for CPU core density.
Intel’s ARK highlights the differences between the processor generations.
Conclusion
DigitalOcean is apparently maximizing profit by increasing density while reducing per core performance on old hardware. Perhaps the servers sold or leased to them at a discount? I don’t know, but certainly doesn’t look like much of a step forward (if at all) in terms of performance.
But now we know.
Comments