Designing Infrastructure for SAP Migration: Important considerations

If you’re moving your SAP application’s home to the cloud, you need to make sure it’s designed for your happy family. Here are some considerations you’d do well to keep in mind.

Moving to the cloud brings several benefits…when done right. The first pit stop on the SAP migration roadmap — which we explore in this post — is designing the infrastructure. In order to handle everything, from baseline requirements, such as performance, scalability and recoverability are both critical and challenging. It is, in fact, a challenge we always welcome at 1CloudHub.

For a business critical application like SAP to function at peak standards, the underlying cloud infrastructure needs to be right-sized, resilient, scalable, elastic and highly available. Achieving this needs a meticulous and thoughtful design.

Infrastructure design needs more focus and care than is usually given to it. It must encompass network design; connectivity design; security and governance aspects of infra design; server and storage design; and as-applicable high availability (HA) + disaster recovery (DR) design; and the underlying OS system compatibility in cloud and SAP database migration design (either HANA or non-HANA).

To create the right infrastructure for SAP migration to the cloud, design efforts need to consider SAPS and memory requirements, IOPS requirements, bandwidth requirements, availability requirements, RPO/RTO requirements, performance, security requirements and including any transition state requirements. In this post, we’ll take you through things we consider thoroughly before we begin designing SAP migration to the cloud.

Right-sizing (Compute, Memory, Storage & OS)

This is critical for two reasons — to be cost-efficient and to match SAPS requirements for ensuring performance. Each cloud provider has multiple instance types benchmarked and certified by SAP for its SAPS. Choosing the right cloud instance from the various options is critical — either finding a balance between compute and memory capabilities for the application server or being memory intensive for the database server.

Insufficient RAM allocations can be fatal for your database — it can cause user experience issues and could end up also generating excessive I/O causing significant performance degradations and increasing storage I/O load. Therefore, choosing the right virtual machine (VM) based on the recommendations of SAPS — SAP application performance standard; wait, shouldn’t it be SAPAPS, never mind! — and memory requirement is crucial.

Choosing the right storage volume based on IOPS is also important. Cloud platforms offer various storage types such as those with assured IOPS, burstable IOPS with a cap, high throughput for serial access and those for random access. Each of them is priced differently. It is essential to ensure we don’t overprovision the space as even a relatively small overprovisioning could cumulatively affect cost incurred. Remember: On the cloud, it is much simpler to size up your storage than to size down your storage!

Our next step is to choose the right OS version, whether Windows, RHEL (Red Hat Enterprise Linux) or SUSE (depends on factors like familiarity, simplicity, server requirement and stability. While RHEL has all the tools – corporate supportprofessional certificationhardware certification, and automatic online updates with Red Hat Network (RHN) — that keeps CFOs and CIOs happy; SUSE is designed for business use and is enterprise-ready from installation, making it easy to work with a variety of office programs. It’s flexible enough to run on many devices but reliable enough for extremely critical processes. Those with Windows familiarity will find it convenient to continue with the same, especially given they’d have the skill to maintain it in-house.

Each of these provides key features of OS level clustering which plays an important role in meeting the availability requirements on the cloud. Also, key to consider is cost and comfort of OS licensing — including portability and support. There are BYOL (hybrid models) to BYOS models available on Azure.

For any other OS (non x86 OS), re-platforming is crucial for implementation on the cloud. Thankfully SAP native methodology in re-platforming helps in ensuring data integrity aspects of migration.

Connectivity, High Availability (HA) and Disaster Recovery (DR)

An essential parameter of infrastructure design is aligning HA/DR Strategy to include clustering, load balancing, DB data sync, shared file storage (NFS, SOFS). In addition, DR is a key aspect at least to a backup-retention-restore level, depending on RPO and RTO requirements of the business application.

Often this configuration takes the form of implementing highly available clustered servers for an application within a production datacenter and having backup hardware in the recovery datacenter. With data from the production server backed up or replicated to the recovery data center, systems are both protected from component failures at the production data center and can be recovered during a disaster at the recovery data center.

The ultimate combination of high availability and disaster recovery occurs when servers are configured as “active-active” or across geographically diverse data centers. In this case, clustered servers for an IT application reside in two different data centers connected by a load balancer and a very low latency data connection. Data between the two servers is synchronously replicated and both systems are “active” at the same time. Should one datacenter be impacted by a disaster, the server in the second internet datacenter picks up the full load of the application and continues uninterrupted.

These are achieved in the cloud by leveraging the concept of Availability Zones in the Infrastructure Design. And with OS level clustering like WindowsFailoverCluster (WFC) or Pacemaker clusters in addition to Load Balancer or DNS routing to achieve the failover for availability. Data synchronization with tools like DRBD or scalable shared storage options of Scale-Out FileServers (SOFS) or NFS on HA Clusters to provide the necessary resilience.

It is also very important to setup resilient connectivity to start with such as a site-to-site VPN connection based on the bandwidth requirement. A site-to-site VPN will allow offices in multiple fixed locations to establish secure connections with each other over a public network such as the Internet. There are hub-and-spoke design patterns available in addressing complex multi-location connectivity possible. As part of the security measure, it is essential to ensure any web traffic which flows via the Internet to be on HTTPS traffic where data sent is secured as it is encrypted and saved.

Another aspect ensuring data security is using ExpressRoute connections or Direct Connect for dedicated bandwidth based on the business requirements. AWS Direct Connect and Azure ExpressRoute let you create private connections between their data centres around the world and your on-premise or hosted infrastructure without going over the public Internet.  This offers more reliability, faster speeds, lower latencies, and higher security than typical connections over the Internet. Resilience can also be built between a dedicated connection to be primary and S2S VPN to be fall back. 

Ramp up on demand and just-in-time

This allows flexibility to ramp up the infrastructure resources required to run SAP on the cloud as per business and project needs. In a greenfield implementation, initially, dev environment is built. It takes a few months before the QA environment is built and eventually the production and sandbox environments. It is easy to upscale or downsize the existing infrastructure as and when the demand comes to meet project specific needs. With the cloud, this can be done instantly and cost-effectively, paying as per use of resources involved in the project.

If rightly done can lead up to 30% cost saving on the first year, that is during project implementation.

Design effectively & get it right the first time, on time, within budget

A good, thoughtful design paves way for a long and fruitful future on the cloud. 1CloudHub’s design process takes into account not just present needs but accommodates your business ambitions and plans. It offers you the blueprint for growth that is unfettered by your technology infrastructure!

Sharing is caring!

In Blog
Subscribe to our Newsletter1CloudHub