Home > Exchange Server, SQL Server, Storage & Backup > Back to #Storage Basics #Exchange2010 #Sharepoint2010 #SQLServer

Back to #Storage Basics #Exchange2010 #Sharepoint2010 #SQLServer

February 9, 2011

Block & Bytes

Understand and know the difference between Block Level Data and File Level  Data and  its storage….Its kind of key.
Refresh your memory here

Grasp the  Fundamental  Technologies: NAS Vs SAN (or DAS)

To select the most apppriate technologies such as NAS, SAN or DAS, you need to have a grasp of what each offers over the other. And to do this you first needed to have understood the above.  Some people think they need one, when actually other would suffice, or their workload dicates.
Refresh your memory on NAS / SAN here

Don’t over engineer a solution, think elastic, dynamic and above all be agile to changing demands, and workloads.

Think Billy Connelly Buisness Plan, but replace business needs with  Storage Needs….
We Want this (Size/ Scale/ Growth), and that (Speed/IOPS), plenty of this (lower TCO/ ROI), none of that (latency), none of this (Corruption/failures), and ***** stay awake as demands (workloads / app requirements)  will all be changed…..Billy Connelly Buisness Plan

Mind your Langauage

Learn the differences in the Storage Languages. How do the Storage protocols/ Transport methods stack up… FC vs iSCSI

Refresh your memory here.


RAID & Writes

The thing to remember is Reads will be generally consistent across all RAID Levels (1 IOPS per Read). However, Writes, will may be clearly different (4 IOPS per Write for RAID5)
(this is largely (not wholey) due to distributed parity, performance traded off for fault tolerance)

 RAID 1 Write Penalty

 

IO Impact

IO Impact

 

 

RAID 1 Write Penalty

Two Writes must be performed for every Write operation. Increased Read performance.
Two IOPS per Write Operation

The RAID5 Write Penalty

Because RAID 5 has distributed parity, two Reads and two Writes must be performed for every Write operation. Four IOPS per Write Operation

RAID 1+0

Creates a striped set from a series of mirrored drives. Two IOPS per Write Operation.
Half the RAID 5 Write Penalty, increased Reads.

In a failed disk situation, RAID 1+0 performs better because all the remaining disks continue to be used. The array can sustain multiple drive losses so long as no mirror loses all its drives.

RAID 0+1

Creates a second striped set to mirror a primary striped set.

The array continues to operate with one or more drives failed in the same mirror set, but if drives fail on both sides of the mirror the data on the RAID system is lost.

LUN scaling and performance  

LUN performance will vary according to the disk or configuration of disks upon which they reside, so it is important to consider the physical medium and its characteristics when planning LUNs as part of storage provisioning.

For example, a LUN that resides on a Fibre Channel 15,000 rpm disk will perform better than an identical LUN on a 5,400 rpm SATA disk. Raid configuration also affects performance and reliability, so the characteristics of the RAID type used for LUNs need to be taken into account.

Type of Workload

Sequential or Random Read/ Writes all impact the design descisions. DataBases – OLAP vs OLTP for example….Are you more interested in read performance than write performance. Are you willing to trade performance for Fault Tolerance and Data Integrity.
 

Calculate IOPS Required

Great post here by Scott Lowe

Table below courtesy of Scott’s article

Average IOPS / Drive

Average IOPS / Drive

 

The Right Questions to Ask

Before you can propose a solution you have to know the problem. Ensure happy customers by aligning the technology to the requirements
Know what questions to ask?

Read Storage area network fundamentals – Part 2

Read Storage area network fundamentals – Part 3

Advertisements
%d bloggers like this: