I read somewhere that the load from LVS doesnt show up when using "top"
for some reason...
But with a site with a load level slightly lower than that but big image
files - I was seeing zeros across the board in my uptime and top.
# ipvsadm -l -n --stats
IP Virtual Server version 1.0.10 (size=65536)
Prot LocalAddress:Port Conns InPkts OutPkts InBytes
OutBytes
-> RemoteAddress:Port
TCP x.y.z.a:443 54737612 2259M 0 539G 0
-> x.y.z.b:443 332044 13485714 0 4095M 0
-> x.y.z.c:443 259853 13974999 0 4006M 0
-> x.y.z.d:443 273464 13566309 0 3796M 0
-> x.y.z.e:443 268103 13662133 0 3683M 0
-> x.y.z.f:443 254818 13534250 0 3969M 0
# w
02:10:58 up 324 days, 10:31, 1 user, load average: 0.00, 0.00, 0.00
USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
# more /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 2
model name : Intel(R) Pentium(R) 4 CPU 2.80GHz
stepping : 9
cpu MHz : 2795.303
cache size : 512 KB
bogomips : 5583.66
For DR, an Athlon should handle it fine, from what I see.
I'd say Gig netcards, people reccommend them for firewalls etc because
they have more buffer capacity, not because you are pushing Gig speeds.
The boys on the pf list recommend SK's I am looking at the SK dual Gig
cards...
You might try to measure latency, since that is the thing you want to
optimize over other parameters on the directors,
if you are using DR. If you are using NAT the equation changes because
your not just handing off requests.
One test: http://www.accs.com/p_and_p/GigaBit/conclusion.html
YMMV,
Rob
mike wrote:
I'm looking into using LVS for load balancing.
I'm thinking I will be using the keepalived implementation and use the
direct routing (LVS-DR) method for forwarding (just for some
background...)
I want to pick the hardware with this in mind; will LVS benefit most
from more RAM? more CPU? what things should I consider (I'm a newbie
at this type of service, so I'd like to ask the people using it what
seems to be the best approach)
Would a single Athlon 64 suffice? or Opteron? 512MB RAM? 2 gigs? Dual
processors?
I'd be looking for the "ideal" situation, but if it could be ordered
based on the pressure points, that'd help a lot. i.e. if processor
cache size really matters, clock speed, memory speed, amount,
configuration, etc.
Right now, over web alone, it appears I'm doing 12-15 million requests
per day. I expect this number to grow a lot more. I currently have 6
webservers serving up content behind my existing [blackbox/ISP
provided] load balancing service. I want plenty of headroom to grow.
I do intend on building 2 identical machines so I have a slave
director for redundancy, as well.
Thanks in advance.
- mike
|