Not logged in » Login
Jun 24 2020

Storage scaling: Vertical and horizontal scaling options with Fujitsu ETERNUS

Storage scaling: Scale-up and scale-out options available with Fujitsu ETERNUS

The purchase of a storage array does not end with zeroing down the vendor/array, capacity sizing, additional features to be invested in and service warranty package taken with the product. One of the major aspects of post-sales purchase is investment in additional storage capacity, which could range from a few TBs to ZBs. This is also referred to as scalability, which in a generic sense revolves around adding more capacity to a storage array in the form of more drives or larger-capacity drives.  

However, given the current data center norms, the concept of scaling the storage arrays can no longer be restricted to just capacity expansions, but should also take into account other factors, like the hosts and network infrastructure involved, applications and workloads supported by the storage array and their subjected requirements, among other things. Doing the proper math on all of these aspects is important, as scaling an array in the form of adding more drives might look like a simple task, but things can get complicated if throughput or performance needs are not considered, which can even jeopardize the entire investment.

 

Things to keep in mind while scaling storage array:

  • Physical or virtual expansion: If scalability beyond a physical array level is sought, then admins can opt for virtual expansion. This is again dependent on the array, and it makes most sense where data center is built on heterogeneous components;
  • Protocols for scaling: If the addition or scaling of storage has an impact on the hosts, the network fabric, the nodes in cluster or other components, preparation for the same will be required;
  • Understanding actual capacity requirements: How capacity is calculated might differ with different vendors and product lines offered. Data reduction features (provisioning, deduplication and compression) might work differently in different cases (inline in some), and they can be switched on/off, or always switched on, offering different ratios of data efficiency, per application. Data reduction can also play a role in how data is replicated across sites in a DR site or cluster scenario. All these aspects will have to be considered when it comes to capacity planning for storage array, not only today but also in the future;
  • Understanding application requirements: There could be some applications which need not only horizontal, but also vertical room to grow over others. This may happen in the form of capacity as well as performance;
  • Scalability parameters to be considered: Addition of more drives, ports, expansion enclosures, system cache, additional cabling; the maximum that the respective array can handle should be taken into account, otherwise resource contention issues will be experienced.

 

Types of storage scaling: Scale-up and scale-out

Scale-up

Initial storage purchase can be further enhanced with the addition of  expansion units (as required) with the desired number drives for capacity, RAM, other software licenses for operations, even controllers, post purchase.

This is a vertical expansion process which is more traditional as well as tried and tested. It makes sense when the organization or admins are looking to maximize the existing hardware. It is relatively simple and easy to accomplish, and far easier to manage since it also brings with it a level of familiarity. Also, it does not call for heavy investment from a budgetary perspective.

 

Image

Scale-out

Initial storage purchase can be accompanied by the addition of new hardware to the existing physical units.

This is a more horizontal expansion, where the limitations of expansion for the physical base unit (initial storage purchase) can be transcended. New hardware will bring better and more improved features. It will provide a new life cycle for existing hardware. Though it might be little costly from investment point of view, the actual return of investment can definitely be derived over a period of time.

Image            

Which scaling method makes sense where?

For simple expansions, a scale-up process makes most sense. One thing though, expansions on the array are dependent on the base controllers as everything is processed via them. As for the unit as a whole, when more expansion units, capacity, RAM or software licenses are added, it leads to progression towards the maximum that whole physical unit can handle. Thus, for cases where capacity in ZBs are sought over the life cycle of the product, a scale-up approach might not be the best fit. Having said that, the scale-up storage arrays available today in all-flash or hybrid variants are more than capable of exceeding the delivery model of what customer’s applications might seek from them.

For data center environments which are built on heterogeneous components and are looking at unprecedented data growth (from TBs to ZBs), a scale-out storage will make most sense. For in this case, more than existing and to-be-added hardware, the magic lies with the software, now that could be the operating system for the storage array or additional software which will manage the entire set up. For maxing out the life of storage hardware also, this method makes sense as storage workloads can be distributed across multiple storage nodes and not just different types of drives or RAID groups. As and when more capacity and performance is needed, more clustered nodes with more capacity can be added. They will also bring more memory, ports, bandwidth etc. apart from more processing power. The only points to be considered while investing in scale-out models are capacity planning and provisioning during initial stages and how system management and upgrades will be handled at a later point of time.

Customers who would want to start small and are looking at massive data growth at a later stage or stages perhaps should consider investing in a storage scale-up model that can also scale-out, if required.

 

An option for every storage scaling need with Fujitsu ETERNUS

We understand different requirements will have to be addressed differently and that is why, with the ETERNUS family, Fujitsu has a solution for every storage-related need of the customer.

 

Image

 

Staring with ETERNUS DX60 S5, we are looking at a model in the economy segment which can handle every basic storage need of the customer. Next is ETERNUS DX100 S5 which offers 4.4 PB of raw capacity and can house both file and block level data. This is followed by ETERNUS DX200 S5, which is regarded as one of the best models in the industry in this segment and can offer 8 PB of raw capacity on 264 drives. ETERNUS DX500 S5 and DX600 S5 are models in the midrange segment which can scale up to 17.6 PB (raw) and 32.4 PB (raw) respectively. They do come with extreme cache (dual port NVMe SSD) for accelerated performance boost. Next is ETERNUS DX900 S5, which can again scale up to 70.7 PB (raw) and comes bundled with NVMe SSDs as additional cache. Followed by ETERNUS DX8900 S4 which is a model which can scale-up to 24 controllers, can handle up to 141 PB (raw) capacity and can deliver up to 10Mn IOPs. These models together constitute the hybrid side of ETERNUS Primary Storage family.

With ETERNUS AF150 S3 we have one of the only models of its kind in this category, it offers 92TB (raw) on 24 drives and can deliver up to 390K IOPs. Followed by ETERNUS AF250 S3 which can handle up to 8.1 PB (raw) capacity and deliver up to 600K IOPs in the entry all-flash segment. Next in line is ETERNUS AF650 S3 which can handle up to 32 PB (raw) capacity and can deliver up to 920K IOPs in mid-range all-flash segment. These models together constitute the all-flash side of ETERNUS Primary Storage family.

To find out more about Fujitsu ETERNUS Primary Storage arrays, please visit this page.

 

Megha Shukla

 

About the Author:

Megha Shukla
close share field

Megha is part of the Product Marketing team, Fujitsu Global business. She has several years of work experience with multiple MNC's across Enterprise IT space...

SHARE

Comments on this article

No comments yet.

Please Login to leave a comment.

X

Please login

Please log in with your Fujitsu Partner Account.

Login


» Forgot password

Register now

If you do not have a Fujitsu Partner Account, please register for a new account.

» Register now