Title: | ase |
Moderator: | SMURF::GROSSO |
Created: | Thu Jul 29 1993 |
Last Modified: | Fri Jun 06 1997 |
Last Successful Update: | Fri Jun 06 1997 |
Number of topics: | 2114 |
Total number of notes: | 7347 |
Hi, We are competing against PDB of SUN very hard to win. We proposal ALPHA 8200(2CPU,1M), OPS, Trucluster. [CLientA] [ClientB] [CLientC] | | | []-----------------------------------------[] Server #1 --------- MC1 --------- | |===================| | | | MC2 | | | |===================| | | | | | --------- -------- | | /test(virtual ip=1) In above configuration, Customer want to use MC2 only for NFS Service and MC1 for Disk Service,Login Service and so on. Because most of the job from client is open,read,write of file which is IMAGE FILE so customer don,t want to Only one MC. Customer think if use just one MC for all Services, MC has so much overhead. Anyway, Customer want to dedicate MC2 for NFS Service.It,s critical and key point to win. We must suggest satisfied result. Question> 1.Can we use MC2 only for NFS Service? If possible, explain more datail and how to get material? If impossible, Waht is the another solution to improve NFS Service? After posting, I will use this note to persuade my customer tomorrow morning.
T.R | Title | User | Personal Name | Date | Lines |
---|---|---|---|---|---|
2025.1 | Only one active MC | NNTPD::"[email protected]" | Dave Cherkus | Mon Apr 28 1997 10:21 | 11 |
The production server product only supports one active memory channel board. It has a mode where a second board serves as a hot standby, but again only one board is active at a time. Same is true for the memory channel driver product. You should talk to product management about future product directions. I am not able to answer your question about the optimal configuration for NFS. I have little experience using NFS in the cluster environment. Dave [Posted by WWW Notes gateway] | |||||
2025.2 | KITCHE::schott | Eric R. Schott USG Product Management | Mon Apr 28 1997 10:46 | 64 | |
Comments enclosed > > Hi, > > We are competing against PDB of SUN very hard to win. > > We proposal ALPHA 8200(2CPU,1M), OPS, Trucluster. > > > [CLientA] [ClientB] [CLientC] > | | | > []-----------------------------------------[] > > Server #1 > --------- MC1 --------- > | |===================| | > | | MC2 | | > | |===================| | > | | | | > --------- -------- > | > | > /test(virtual ip=1) > >In above configuration, Customer want to use MC2 only for NFS Service >and MC1 for Disk Service,Login Service and so on. >Because most of the job from client is open,read,write of file which is >IMAGE FILE so customer don,t want to Only one MC. >Customer think if use just one MC for all Services, MC has so much overhead. >Anyway, Customer want to dedicate MC2 for NFS Service.It,s critical and key >point to win. >We must suggest satisfied result. I think the customer has a big miss understanding...the overhead of MC is not very high, and a single MC will handle the load they are placing on the system. The second MC is for redundancy in the event of a problem. We plan to support in a future release, load balancing across multiple MC. While this will provide potential greater bandwidth, for the use your customer describes I don't see them being able to tell the difference. > >Question> 1.Can we use MC2 only for NFS Service? > If possible, explain more datail and how to get material? > > If impossible, Waht is the another solution to improve NFS > Service? I think you need to be clear on what you are trying to improve? What data do you or the customer have the MC is a bottleneck in what they are doing? If it were to be found to be a bottleneck, the future loadbalancing would help with this...I don't think you will find it to be a bottleneck. > >After posting, I will use this note to persuade my customer tomorrow morning. > > > |