Case Study

IBS Private Cloud SAN Storage Upgrade

Overview

This is an IBS internal case study about a recent (2017) IBS Cloud infrastructure upgrade project.

Regarding the virtualization and flash technologies in storage area, we consider that our conclusions and analysis of the final result will prove extremely helpful for any company in the SMB and Enterprise segments, when they have to decide on SAN storage infrastructure.

The goal of our project was to modernize the storage layer of our cloud in order to provide better services for our internal and external customers. We aimed to provide better performance and higher capacity.

After successfully upgrading our SAN storage systems we manage to achieve more than 3 times higher performance with significantly lower operating costs

Introduction to the environment

Our Cloud serves around 650 virtual machines with a mix of different loads. It provides virtual machines for our Software Development group, runs our ERP, CRM and e-commerce systems, Enterprise Data Warehouse with Cognos Business Intelligence, Virtual Desktop Infrastructure, our Cloud services – Busticket.bg, Timeoff.Guru, Bus Ticket Management system, Theater Management System and others. It also provides dedicated virtual machines to our hosted customer environments.

The original core data layer was built on two EMC storage servers - Clariion CX4-120 with a mix of FC and SATA disks with RAW capacity of 45TB (35TB after RAID).  Over 80% of the capacity was utilized and in peak times our combined performance was 25000 IOPS with latency 4-8 ms (4KB, 70R/30W) at best. We installed this infrastructure 7 years ago and obviously needed to replace it as soon as possible. The amount we invested in 2010 was 59К Euro.

Our Goals

Before selecting any products we defined our technical goals and restrictions for the project:

  • Bring down the latency below 2.5 ms
  • Provide at least 15TB more capacity
  • Ensure no services interruption when moving to a new storage infrastructure.

Platforms selected

Keeping in mind the limitations and goals we selected IBM Storwize Virtualize (SAN Volume Controller) with SSD based all-flash system Storwize 5030F for our data backbone layer.

IBM SVC was a key strategic decision. It gives us freedom to select any storage array from the market as backend storage at any point in time and guarantees smooth migrations with no interruptions. It provides essential additional benefits for us like:

  • Single SAN management platform to operate with
  • Advanced SAN functionality – flash copy, mirror – independent from storage array vendor or model
  • Hardware based compression in real time

Since we had a limited budget - the mid-range model all-flash system Storwize 5030F, configured with the Read-Intensive SSD media was the obvious choice. The installed capacity of 35TB after RAID 6 is built with 3.84GB RI SSD.

35TB only? What about those 15TB additional capacity?

Remember IBM SVC hardware compression – it saved us more than 48% storage space. This resulted in 13.44TB savings from our current data. And considering that at average any new data will be 48% compressed we end up with more than 20TB free space – equal to about 30TB on our old infrastructure.

 

Budget

The new system cost us 82K USD with extended warranty to 5 years. This is about 66K Euro at the time of acquisition (2017), which is 10% more than the budget 7 years ago. We considered this acceptable after calculating the TCO and the projected savings (look at the last paragraph)

 

Performance

We achieved 75000 IOPS, which is exactly 3 times more compared to the old system.

We had some doubts whether we will reach our goal for latency, but even with hardware compression it is down to 1.5-0.5 ms.

 

Migration

The entire process took less than 3 working days. There was no night or weekend shifts and it lasted 3 days because we were extremely careful and tested each configuration 3 times and more before execution. No user or process was affected or experienced anything unusual just the effects of the 3 times increased performance. VDI users were the first that noticed that something is changed for better.

 

What about TCO?

The new installed system fits in 6U in comparison to 18U for the old system, which being combined with dramatic power savings from SSD technology resulted in 400 Euro per month savings for our Cloud installation. For the next 5 years we will save 24 thousand Euro just from that. When we add the failed disks expenses, we previously had – 2.8K Euro per year (average), our savings for five years will sum up to 38K Euro, without calculating any additional benefits from simplified systems management.

 

At IBS we recognize good reliable technology. We do SAN infrastructures for more than a decade. To IBM we want to say just “Thank You! Thanks for the technology that just works.”

 

If you want to learn more about different flash storage systems visit our dedicated web page www.ibs.bg/flash (bg)

You can learn more about our services for storage at https://ibs.bg/wps/portal/ibs/products/it-infrastructure/storage

 

 

 

 

Keywords:
case-study,it-infrastructure,storage
Share it: