Storage Boards Cleanup
To make it easier to find information about HPE Storage products and solutions, we are doing spring cleaning. This includes consolidation of some older boards, and a simpler structure that more accurately reflects how people use HPE Storage.
HPE StoreVirtual Storage / LeftHand
cancel
Showing results for 
Search instead for 
Did you mean: 

ALB versus NLB on DSM 4000

SOLVED
Go to solution
ictgvbsb
Frequent Advisor

ALB versus NLB on DSM 4000

The official documentation of HP is not clear about the working of ALB.

 

What I see is that on specific server both NIC`s are used ans other servers only one NIC.

And al the servers are configured with HP DSM 4000 drivers.

 

Can someony exlpain this behaviour?

 

 

5 REPLIES
RemyZ
Advisor

Re: ALB versus NLB on DSM 4000

We struggled with this as well. Since P4500 is Linux based, this URL explains the working of ALB:

 

http://www.mjmwired.net/kernel/Documentation/networking/bonding.txt#529

 

--------------------------------------
Remy Zandwijk
VU University Amsterdam
ictgvbsb
Frequent Advisor

Re: ALB versus NLB on DSM 4000

Hello remy,

This is indeed a detailed discription of ALB on working on the side of the Lefthand box.

But what I see is strange behaviour on my Windows servers. Somtimes there is load on both ISCSI interfaces but manly

DSM 4000 driver usses only one NIC. The driver is set to the default Vendor specific mode.

 

 

Cajuntank MS
Valued Contributor

Re: ALB versus NLB on DSM 4000

You have to manually create the iSCSI connections via both server NICs and when that's done, change the MPIO from vendor specific to round robin. This info is assuming you are using HP DSM for MPIO.

ictgvbsb
Frequent Advisor

Re: ALB versus NLB on DSM 4000

Hi there,

 

Ik know how youy can set it. But I want to know what is best practise. With the vendow specific setting I see different behaviour.

 

Some servers use Both NIC`s for 30 and 50 %

Other servers use a singel NIC 50 % and the other at top 0.9%

 

Iam investigating this because my SQL server only gets with SQLiO meter 1800 IOPS on a 10 node Lefthand cluster

ictgvbsb
Frequent Advisor
Solution

Re: ALB versus NLB on DSM 4000

when I exactly follow the manual of

 

http://www.brentozar.com/archive/2008/09/finding-your-san-bottlenecks-with-sqlio/

 

The performance comes out just fine. Although there is only one NIC  used of the two by HP  DSM4000 ALB driver the IOPS are great. SQL uses 8 KB blocks. YU yust need to run the test with the right parameters!

 

C:\Program Files (x86)\SQLIO>sqlio -kW -t8 -s120 -o8 -frandom -b8 -BH -LS I:\tes tfile.dat sqlio v1.5.SG using system counter for latency timings, 2864628 counts per second 8 threads writing for 120 secs to file I:\testfile.dat         using 8KB random IOs         enabling multiple I/Os per thread with 8 outstanding         buffering set to use hardware disk cache (but not file cache) using current size: 24576 MB for file: I:\testfile.dat initialization done CUMULATIVE DATA: throughput metrics: IOs/sec:  7557.70 MBs/sec   59.04 latency metrics: Min_Latency(ms): 0 Avg_Latency(ms): 8 Max_Latency(ms): 1046 histogram: ms: 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+ %:  1 52  6  3 18  1  0  0  0  0  0  0  0  0  0  0  0  0  0  0  1  1  0  0 16  

 

 C:\Program Files (x86)\SQLIO>sqlio -kR -t8 -s120 -o8 -frandom -b8 -BH -LS I:\tes tfile.dat sqlio v1.5.SG using system counter for latency timings, 2864628 counts per second 8 threads reading for 120 secs from file I:\testfile.dat         using 8KB random IOs         enabling multiple I/Os per thread with 8 outstanding         buffering set to use hardware disk cache (but not file cache) using current size: 24576 MB for file: I:\testfile.dat initialization done CUMULATIVE DATA: throughput metrics: IOs/sec:  8881.25 MBs/sec:    69.38 latency metrics: Min_Latency(ms): 0 Avg_Latency(ms): 6 Max_Latency(ms): 742 histogram: ms: 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+ %:  1  2  8 12 15 15 11  8  6  5  4  3  2  2  1  1  1  1  1  0  0  0  0  0  1