How to maximize effectiveness with Hybrid Storage?

Introduction

Nowadays, the demands for storage’s high performance with low latency increase than before. Traditional disks can no longer satisfy the requirement in the storage markets. Fortunately, SSD came out with lower and lower price. Users have a better choice to meet their challenge in the IT environment. QSAN provides two features, SSD cache, and Auto-Tiering, both functions can provide robust performance with SSD plus traditional disk. But you might curious about what is the difference between SSD cache and Auto-Tiering feature. And what’re the benefits that they can bring up to you. We will compare both features in this blog as below.

 

Overview

We all know AFA(All-flash array) can satisfy for all the user who has the high demand for performance and low latency. However, the AFA budget is higher than traditional disk array. And the lower capacity of SSD will also be another issue for the user. Here, we will introduce SSD cache and Auto Tiering. Both can conquer the budget and capacity issue and bring powerful performance as the same as AFA array.

 

SSD cache

QSAN SSD cache, you can consider it is a secondary cache which consists of SSDs between controller’s memory and hard disks. By our SSD cache mechanism, if the IO structure of the user’s environment is all random read/write type, SSD cache can improve the IOPS remarkably.
Download SSD Cache 2.0 White Paper >

Benefits – Read Cache

With SSD read cache technique, random read access triggered from the connected host(s) will get higher IOPS. And lower response time whenever the read request has been cache-hit. It means the user can read data from the SSD cache pool instead of the traditional disks in the RAID pool. It will deliver high IOPS like what AFA may achieve. Though this is only beneficial with read access and random I/O.

Benefits – Read Write Cache

Same as SSD read cache case; higher performance with low latency can be obtained once the read/write I/O is random. And it will be written into the SSD cache pool, high IOPS like what AFA can achieve will be delivered with the lower budget. Meanwhile, you may keep the high total capacity from the traditional disks in the RAID pool.

SSD cache function needs to put the hot data (frequently accessed data) into the SSD cache pool before it can deliver the high IOPS value. It requires some time to warm up the data into the SSD cache pool after the system’s calculation for the incoming/outgoing I/O from time to time. And the overall performance will get a bit lower than usual before the warm-up time finishes.

 

Auto-Tiering

QSAN Auto-Tiering is the pool base feature. There are three tiers(disk groups) consisted of the Auto-Tiering pool which are the high tier – SSD, medium tier – 15K SAS, and low tier– NL-SAS. According to our Auto-Tiering mechanism – data analyze and Auto-Tiering policy. The User can make the data be read/written by a specific tier and make the data be moved between each tier. You can refer more information about Auto-Tiering as the below link.
Download Auto Tiering 2.0 White Paper >

Benefits – Auto Tiering

Auto Tiering function has the ability to achieve what SSD cache can do, and not limit in random I/O only. With the five policies mentioned in the white paper above. You may specify a volume to be written with data into the highest tier all the time. You may also dynamically change the location/tier of the written data afterward by changing the policy to be the lowest tier for the specific volume.

This kind of flexibility gives the IT manager an ability to move the volume to the highest tier (SSD disk group) in time. Imagining that the volume is assigned to a Dept. if the Dept. requires to gain higher performance than usual for a short period. IT manager can change the policy of this volume to the highest tier, and relocate the data. The Dept. will get the fastest performance and response time just like what AFA array can do. Relocating the data back to the lower tier if necessary afterward. Users won’t need to purchase AFA array in such circumstances, this kind of Hybrid storage with QSAN Auto Tiering function can give users what they need. Besides that, more benefits are:

  • Flexible Configurations
  • Optimal Storage Efficiency
  • Cost Saving
  • Schedule Relocation
  • High performance
  • Tiering Analysis

 

How to choose

According to the introduction of SSD cache and Auto-Tiering, you might curious about how to choose and which feature is suitable for your environment. We will guide how to choose in the below conditions.

  1. Realize how much SSD cache capacity will need to use for hot data
  2. 2. Make sure the I/O structure of the user’s environment is full random read/write type
  3. User’s storage environment is more like small enterprise, storage quantity is around 1 ~ 3 only

If your environment doesn’t conform to the above conditions, we will suggest considering to choose our Auto-Tiering, it should be more suitable for the large enterprise environment.

 

Case study

Let’s take a case study as a reference. If the user who enables the SSD read/write cache with 500G SSD capacity, and the user doesn’t know how much SSD capacity he needs. Then the user will gain faster performance with random read/write environment only before the SSD capacity running out. As long as the random data grows up near to 500GB, SSD cache capacity is almost running out, and the system will start to flush the cached data back to the RAID pool. Performance/IOPS will then drop to be lower than the scenario of SSD cache disabled condition because the traditional disks in the RAID pool need to absorb the front-end I/O as well as digesting the data flushing from the SSD cache pool.

In this kind of scenario, it is suggested to use Auto-Tiering. But if the user realizes the random data range, it is doable to create a small volume with capacity lower than the SSD cache pool size, using this small volume to be created as a Datastore where a VMware environment uses to put booting data of many Virtual Machines. It is much suitable by using SSD caching function.

 

Conclusion

With both features, QSAN provides the user with more suitable solution of their environment. The cost will be lower than AFA(All flash array), make the budget won’t be the major concern of all user. Simultaneously, the user can get the similar performance with AFA, but the capacity is larger than AFA undoubtedly. How about trying it to see the overview of SSD Cache and Auto Tiering with our Demo unit now?