Thursday, July 11, 2019

Storage Pools: Homogeneous vs Heterogeneous

 Homogeneous in storage system means using identical disk in a single array, and identical itself defined same characteristics: disk type (SATA, SAS, Flash), throughput (I/O Speed), capacity (thousands of GB or TB or even more), vendor (EMC, HP, NetApp) and etc. In contrast with Homogeneous, Heterogeneous simply means mixture and complexity of array of disks.
 Although Storage Tiering are executed based on required throughput and functionality and also total price of disk and Service Level Agreement and made it possible for storage pools to use variety of  But what is the best option? is there any certain decision to choose mixture (Heterogeneous) or simplicity (Homogeneous)? There's some benefit in Heterogeneous in enterprise scales, because we can create pools with different disk types and different level of storing services. A big problem is how it's possible to transfer data between different tiers of disk when it's required without any interruption or required manual action for data migration between Homogeneous storage pools.
  EMC provides a good answer for this situation: FAST VP. This great feature offer the highest flexibility that storage device can reached by allocating storage pool in different tiers based on performance requirement. Remember that tiers consist different types of disk but it's strongly recommended to use the same speed in each type, not rotational. Then storage system makes decisions about the best place for a data based on its importance, needed throughput. Heterogeneous pools are  fundamental storage infrastructure for EMC FAST VP. (Note that Thick LUN is suitable as FAST VP underlay)
 About the applications and services with similar requirements, Homogeneous Pool (Tier with single type of disks) is better option because there is an expected Input/Output stream of bandwidth and data storing perform, So there is no needs to tiering data between different storage types. Heterogeneous pools can consist different types of drives. As Tomek said in this post FAST VP facilitates automatic data movement to appropriate drive tiers depending on the I/O activity for that data. As a result NL-SAS or even SATA disk types is better for data with low rate I/O or high required capacity (like Backup files) and Flash-based disk (like SSD) is more suitable for low capacity or Highest required I/O (So probably more important data) and at last SAS disks are the best candidate for mid-range of Storage Infrastructure of data (medium capacity density and medium throughput of data). EMC called each of previously mentioned Respectively: Capacity Tier, Extreme Performance Tier and Performance Tier.
 In other view we can say Homogeneous Pool is a good choice for predictable data (respect to each rate: capacity or performance) because you know exactly what you expected from your storage underlay. So each of these pools can select a single type of drive unlike the Heterogeneous Pools. But if I want to generally speaking about every storage products from any vendor without considering any advanced features, sometimes you may prefer the simplest way: create homogeneous pools, assign LUNs to the storage initiators and then in a duration of time calculate the performance metrics and storage space. Then you can say what type of disks with which speed and capacity you really required. So after a time if you need to expand your storage infrastructure
in a complex environment, it may be used Heterogeneous pool.



No comments:

Post a Comment

I will start a new journey soon ...