HP StorageWorks Enterprise Virtual Array Cluster FIO Starter Kit Podręcznik Użytkownika Pobierz pdf (Strona 122)

from the EVA to the DPMs. In that case, 16 back-end volumes would be the recommended minimum

number of volumes for the pool, while 32 back-end volumes would be even better.

• Larger numbers of volumes for a concatenated pool have the benefit described above of providing

more opportunities to distribute the workload across the multiple array ports; however, there are

trade-offs involved. There is a maximum number of 1024 back-end volumes supported per domain.

There is also a maximum number of 4096 paths supported. Pools with larger numbers of back-

end volumes will consume more back-end virtual disks and more paths, and this may result is

running out of back-end volumes and paths before achieving the necessary back-end capacity.

• Note that front-end virtual disks are allocated from capacity on back-end volumes using an algorithm

that roughly distributes the front-end virtual disks across the multiple back-end volumes. This also

distributes the workload of the front-end virtual disks roughly across the multiple paths. (The algorithm

was based on the assumption that pools will be created with at least 8–16 back-end LUNs (which

seems like a reasonable assumption over the lifetime of a system). Another consideration was to

avoid going over all the back-end LUNs in the pool to find the best match (for example, the smallest

contiguous free area that is greater than or equal to the required capacity) in order to save time

(at least for PiT expansions time is a critical factor). When additional virtual disks are configured,

a back-end volume is selected at random, and the algorithm seeks to find a contiguous free space.

This applies to the temporary volumes used to support PiTs, snapshots, thin provisioning, and so

on, too.

• One tradeoff of this simple general purpose pool construction approach is that it uses back-end

volumes and back-end paths in quantities that can run out before achieving the maximum capacity

or maximum number of arrays. The maximum number of back-end volumes supported is 1024.

The maximum number of back-end paths supported to each back-end volume is eight. The maximum

number of back-end paths per DPM is 4K (these correspond to a data structure called physical

storage containers (PSCs).

Building storage pools using stripe sets

The capacity-optimized pools described above are a good starting point, but it may not be the best

choice for all situations. One issue is that the use of the multiple paths to the back-end volumes is “hit

or miss.” It is usually much better than a construction that uses a small number of paths, but the use

of the multiple paths is not deterministic. Storage pools that are built using stripe sets spread the I/O

for a single virtual disk across the many paths and back-end volumes used to build these performance

pools. As a result, a front-end virtual disk that is created from a pool built with stripe sets will be

sequentially spread across all the members of the stripe set in 1-MB chunks.

Build a stripe set using the same guidelines as those for building a general purpose or capacity-based

storage pool. That is, the volumes should be of the same RAID type, with similar capacity and

performance characteristics. The size of a stripe set is based on the size of its smallest member times

the number of members. This means that unless all members are of exactly the same size there will

be capacity that is not accessible. The stripe sets should be constructed of at least as many volumes

as there are paths from a single DPM to the array. One or more stripe sets are used to create the

pool.

Stripe sets can both improve performance or degrade performance so it is important to understand

the following guidelines:

• Sequential I/O or transactional I/O—If the I/O stream is largely sequential, and the array is able

to detect a sequential stream, it may make more sense to use a pool over a stripe set. Additionally,

if the I/O size is greater than one megabyte, striping it would create two requests. If the I/O

stream is largely transactional, a stripe set will probably perform better.

• The EVA Cluster does not allow another back-end LU to be added to a stripe set.

• The EVA Cluster does allow the addition of a stripe set to a pool, but the EVA Cluster will not re-

balance. A best practice in this area is to add capacity to the HP EVA and present additional

volumes to the EVA Cluster, because the HP EVA will rebalance.

Configuration best practices122

1 2 ... 117 118 119 120 121 122 123 124 125 126 127 ... 207 208

HP StorageWorks Enterprise Virtual Array Cluster FIO Starter Kit Podręcznik Użytkownika Strona 122

Komentarze do niniejszej Instrukcji

HP StorageWorks Enterprise Virtual Array Cluster FIO Starter Kit Podręcznik Użytkownika Strona 122

Komentarze do niniejszej Instrukcji

Powiązane produkty i podręczniki dla Części komputerowe HP StorageWorks Enterprise Virtual Array Cluster FIO Starter Kit