2013-05-31

When it comes to storage management there is a high probability of developing performance issues over time. There can be bottlenecks and hotspots in many places on your storage environment including the storage arrays, the controllers and the disk drives. If the server environment is virtualized, it will lead to more storage performance challenges due over-commitment of resources, abstraction and VM sprawl issue making it difficult to diagnose IOPS spikes and contention given the dynamic nature of
virtualization.

There are various components in the storage landscape that affect the performance of the storage system. These include RAID levels, the number and type of disk drives in a RAID set or volume group, the type of drives and their performance capabilities, as well as host server front-end ports and back-end device ports.

Problems are aplenty: How Can Storage and Virtual Admins Address Them?

Here are 5 simple tips to help address storage I/O performance issues. These are some simple administrative actions that we can perform to avoid common performance issues.

#1 Consider Changing RAID Type

RAID has 2 clear benefits: Better performance and higher availability, which means it goes faster and breaks down less often.

Performance is increased because the server has more "spindles" to read from or write to when data is accessed from a drive.

Availability is increased because the RAID controller can recreate lost data from parity information.

You can change the RAID type of the logical disk drives based on the storage data that you have based on your budget. You can use RAID-10 for high performance everywhere, but it may be costlier. If you have budget constraints, you can use RAID-5 for database data volumes, with RAID-1 or RAID-10 used on database log volumes.

#2 Replace Old Disk with SSD

There are many enterprises that are considering replacing existing hard disks with solid-state drives (SSD) or using SSD drives as a large cache for the array. An SSD is one big piece of flash which makes data access almost instantaneous. Without any moving parts, SSDs contribute towards reduced access time, lowered operating temperature and enhanced I/O speed.

SSDs help to expand built-in cache, thus speeding up all I/O requests to the array

SSDs help storage arrays in storage tiering by dynamically moving data between different disks and RAID levels, and thus resulting in improved performance at the pool (RAID group) level.

#3 Reallocate Disk I/O Load

When different workloads share the same disk, it may lead to interference causing the disk head to shuttle back and forth between different locations based on the workload request. Especially in a multi-disk system data read/write can be very time-consuming.

Data reallocation separates interfering workloads by moving one or several of the workloads to other disks, where they will cause less interference and avoid hotspots. Reallocation also helps with defragmenting data by arranging the stored data sequentially so that when performing a read operation, data retrieval is faster.

#4 Upgrade to Larger Cache

The disk cache holds data that has recently been read, and increasing the size of the cache space will lead to improved read/write operations and lesser performance I/O bottlenecks.

#5 Plan Well and Add More Physical Disk to Arrays

This is the inevitable scenario in any storage environment. No matter how many disks you add to expand your storage capacity, it’s eventually going to exhaust even the most generously sized LUNs. You can keep in mind the following best practices when adding more disks to arrays.

Know how many disks an array can hold, and how many disks are actually installed at the moment

Consider IOPS as well as just disk storage capacity

Understand storage content, classify data based on criticality and usage, and choose the storage expansion option accordingly

Instead of adding more disks, you can also choose to upsize existing disks to larger volumes of storage

Ensure you have a storage capacity planning and predictive forecast analysis of how your storage space is used and is evolving so you can add disks only where it’s really required.

While these are all effective tips to improve storage I/O performance, the bigger issue is to monitor the storage environment and identify bottlenecks as they appear so we can avert performance issues. You need to have top down visibility from VMs, to hosts, to network elements, to storage.

SolarWinds Storage Manager monitors storage performance & isolates hotspots in your multi-vendor SAN fabric. Storage Manager allows you to drill down to view total IOPS at the LUN or RAID Group level to see if there is LUN contention or if physical storage is the bottleneck.

Show more