Blog: How the detailed modernization roadmap helped reduce the number of servers in the data center by 96%

A major American provider of global eCommerce solutions was looking for a way to upgrade their extensive data processing system. We provided them with visibility into their system and built cost-effective and highly optimized modernization roadmap using our DBMSys platform. Using our detailed instructions they can substantially cut the upgrade expenses and take advantage of their workloads migration to Azure along with benefits from virtualization and consolidation options.

Customer problem

Our customer has been supporting a number of data centers around the world for a long period of time. Without timely monitoring and updating such systems easily become outdated with a high percentage of servers running on no longer supported versions of database or operation system. Moreover, the server hardware is rapidly aging over the years and the whole system becomes inefficient and overpriced. That’s why they needed to upgrade and consolidate their servers with an opportunity to migrate them to the cloud.

The customer asked us to analyze the existing environment and based on that lay out modernization roadmap showing how they can reduce the costs while enabling new capabilities.

Analysis

Before making optimization decisions, first we had to fully analyze their environment.

We applied our data collecting platform called Database Management Systems (DBMSys) to build a modernization roadmap using gathered information.

DBMSys not only helps with monitoring an environment for capacity needs, but also provides data over time needed to make informed business decisions. For our customer’s system we were gathering:

• Server utilization metrics
• Metrics on hardware and OS performance
• Database metrics
• Network bandwidth.

Basing on this data, we perform analytics to correlate current capacity to future state environment.

We discovered that 89% of their current environment was virtualized. Most of their operating systems were Windows Server 2003 and 2008 and 54% of installations were greater than 10 years’ old.

The database distribution shows that they also tend to be rather outdated. 56% of all databases are SQL Server 2008/R2 with their support ending in two years, and 26% of SQL Server 2000/2005 with already terminated support.

Moreover, we discovered performance issues such as low CPU scores and low memory on some instances.

Consolidation/Virtualization

Virtualization and consolidation techniques could significantly optimize servers’ utilization and therefore lower costs associated with server maintenance. Both approaches aim at the same purpose, but have different realizations. Consolidation implies combining the different workloads within the scope of a single server. In this case, consolidation decreases operational costs by reducing the number of operating environments and thus minimizing their licensing costs.

As opposed to consolidation, virtualization allows running multiple workloads in specially configured virtual machines that run independently on the same physical server. Virtualization can drop the number of physical servers that have to be managed, but the number of operating systems remains the same. In this regard, we can highly increase the utilization of our servers while reducing their number, which means that the owner will spend less on hardware and maintenance. Moreover, virtualized environment lets us manage and monitor the entire virtual infrastructure from a central location and allows us better plan future data center growth.

Possessing DBMSys metrics of the customer’s system, we gave the following recommendations on consolidation & virtualization:

• Consolidation on current hardware would allow them to decommission up to 9% of their machines. That would lead to the reduction of SQL Server Enterprise cores to 6% and therefore would allow our customer to save approximately $85,000.
• The optimal solution is to virtualize the whole customer environment on 3 hosts. This would dramatically reduce the number of hosts by 96%. Compared to the number of current servers we can get 96% hosts reduction. That means the customer would need significantly fewer servers in their data center. In such a way they could save more than $200,000 annually on servers maintenance and hardware cost.

Azure Adoption

When migrating to Azure, we have the following options:

• Migrating existing servers or existing virtual machines to Azure virtual machines
• Re-creating existing workloads without changes in Azure or consolidating them.

We figured out the most optimal Azure configuration based on servers’ utilization metrics collected by DBMSys platform and current Microsoft specifications. We identified underutilized servers as initial candidates for Azure adoption.

Using DBMSys performance metrics we determined on-premise capacity. That served as the data foundation to determine the appropriate Azure environment.

We combined server capacity with Azure pricing to determine the cost of running the environment in the Azure cloud.

We offered our customer two strategies in order to optimize Azure costs:

1. Migration to Azure SQL Database
   • 94% of customer databases had potential for migration to Azure SQL databases as a PaaS strategy
   • Basing on numbers of current databases’ resource consumption we offered to employ Elastic Pools for appropriate databases to optimize their utilization.
2. Consolidation and movement to Azure Virtual Machines
   • It is worth consolidating 75% of their physical servers on several Azure Virtual Machines
   • Our calculations show that consolidated Azure environment is 18% cheaper than non-consolidated.

Conclusions

In addition to building the detailed modernization roadmap, we recommended the upgrade tools and automation techniques that could facilitate upgrade and migration process. We performed this job with a special focus on risks mitigation, increasing the efficiency, and cutting down the costs.

Summing up all modernization findings that we delivered to our customer:

• The majority of observed database environment is SQL Server 2008 or older versions. Therefore our customer should upgrade their system as soon as possible.
• Customer can leverage migrating databases to Azure SQL Databases (PaaS)
• On-premise environment not suitable for Azure adoption should be consolidated and virtualized
• Consolidating the Azure VMs and right-sizing on-premise VMs will optimize utilization and costs

Feel free to contact us to upgrade your system with confidence.