Virtualizing the physical data center, and then migrating it to the cloud can result in significant operational and capex benefits. Small wonder then, that so many enterprises these days are contemplating cloud migration or transforming to cloud. Spurring this physical to virtual to cloud (P2V2C) trend is the fact that a host of cloud migration offerings have been developed, and are available in the market. Here are a few tips that one should consider while transforming or migrating into the cloud.
- Transforming existing physical infrastructure to the cloud is not as easy, as it may initially seem. It’s a journey from physical to virtual data center to operationalizing, and then migrating to the cloud. Abrupt P2V2C transformation is infeasible; rather, cloud migration is a phased process. First, analyze the physical environment; next, virtualize the physical infrastructure (server, network and storage); next, tune the environment for optimum application performance — for example, add more host bus adapter (HBA) or network interface controller (NIC) cards to the physical server. Finally, a cloud management framework can be selected to manage the virtual infrastructure as a cloud.
- In order to analyze the current physical infrastructure and applications, factor the compute (server), network and storage components of the IT infrastructure, along with application requirements. Application owners will demand that the virtualized environment meets the performance requirement of the applications, and that cloud-based x86 architecture does not pose compatibility problems after cloud migration. A sensible approach is to categorize applications into pools according to performance requirements. For example, SAP and Oracle applications can go into a top-level pool, applications such as Exchange will fall into a second-level pool, HR applications into a third pool, and so on.
- Now comes the turn of legacy applications that are not cloud-compatible. There are two possible cloud migration approaches: One, look for a public cloud vendor offering; or, two, get the application rewritten for the cloud. Applications can be written using cloud frameworks such as REST.
Consolidation and virtualization
Once application pooling is complete, and legacy applications are taken care of, the process of consolidation and virtualization of compute, network and storage, in that order, begins.
- Consolidation and virtualization of “compute” involves consolidating the physical servers and migration of the physical machines to virtual machines. Prioritize the virtual machines according to the application resource requirements, using the offerings of hypervisor vendors such as VMware, Microsoft (HyperV) and others. For instance, one can dedicate CPU shares for business-critical applications, separate the production network from the management network, and so on. While moving to the virtual server environment, downtime of one physical server will lead to downtime of each virtual machine running on it, so always use the fault-tolerant and high-availability features offered by the hypervisor vendors.
- Once physical machines are migrated to virtual machines, monitor resource usage and application performance. Tune the hypervisor parameters until all virtual machines meet the application requirements.
- Now move towards network virtualization by creating VLANs for segregating production traffic from other traffic such as backup data traffic. Also, moving to a Fiber Channel over Ethernet (FCoE) infrastructure means less cables to manage, and hence better efficiency. With vendors expecting the underlying connectivity backbone to have a throughput of 10 Gbps, FCoE ensures that the infrastructure is future-ready. On the physical server front, make sure you have redundant NIC cards, or CANs in case of FCoE.
- For the next step in the cloud migration journey, categorize physical storage based on performance. For example, SSD disks will fall into tier 0, FC into tier 1, and so on. Products such as VPLEX help to consolidate the physical storage into pools of storage based on performance. Apply the tiering policies to slot the existing application data in the appropriate tier. Tuning of storage and application performance might be required. For example, a particular application might require a dedicated pool, rather than be clubbed with other applications.
- Now that compute, network and storage have been virtualized, the cloud migration’s destination is within sight. Using cloud management utilities offered by HP, EMC, IBM or other vendors, the virtualized resources can be managed from a single front-end, replete with features such as on-demand service, rapid elasticity and measured service.
- Lastly, for applications sourced from the public cloud, ensure that the cloud service provider is compliant with stipulations such as SAS 70, HIPAA and PCI. Be aware of the modalities involved in moving data back from cloud to internal infrastructure. Also check the uptime promise for the hosted applications and interoperability with other cloud vendors or technologies. Since you have decided to put a valued and regulated asset in the hands of a cloud service provider, do have a reliable third-party audit of the environment after the implementation, to provide additional assurance and evidence that set goals of cloud migration have been achieved.
About the author: Anuj Sharma is an EMC Certified and NetApp accredited professional. Sharma has experience in handling implementation projects related to SAN, NAS and BURA. He also has to his credit several research papers published globally on SAN and BURA technologies.