A recent report describes important considerations regarding enterprise application workloads that chief financial officers should account for when planning cloud computing strategies.
Each enterprise application workload has its own hardware, storage, networking, availability and redundancy requirements, along with different architectural characteristics. Companies need to think about the various workload requirements for current and future applications, and how they will change as user needs evolve, according to the report by Citrix, a cloud company based in Fort Lauderdale, Fla.
“As organizations define their cloud strategies, it is imperative to think about the use cases and application workloads that will run in the cloud, as these will ultimately define your cloud architecture and which solutions to deploy,” the report states.
The report breaks down application workloads into two basic categories: traditional and “cloud native.”
Traditional scale-up application workloads include most of the current enterprise applications that are housed in data centers. SAP ERP, Oracle database applications and Microsoft Exchange fall into this category, and the applications typically have tens of thousands users and hundreds of sessions running concurrently, according to the report.
To achieve scale with a traditional application, the size of the application and the database infrastructure must be increased. That means the architecture of the workloads has only a limited ability to scale, the report states.
Traditionally, enterprise workloads are designed to use reliable hardware — servers and storage that aren’t expected to fail during normal operations, according to the report. They also have sophisticated backup and disaster recovery procedures.
The second category, cloud native scale-out application workloads, includes the applications typically associated with cloud computing: gaming and mobile apps, big data, social apps and batch processing, Citrix writes. They are designed to minimize cost, not to ensure reliability.
Scaling and elasticity for cloud-native workloads comes from the loosely coupled, commodity-grade computing, network and storage nodes, not through buying larger and larger servers and using traditional data center architecture, according to the report. Scaling and elasticity is also accomplished by running multiple applications servers in parallel fashion, as well as liberal caching and replicating data to multiple databases.
“Built for infrastructure that isn’t expected to be resilient, (cloud native) workloads are designed from day one with the intelligence to handle the failure of any given node simply and efficiently,” the report says. “Cloud native workloads assume that the underlying infrastructure can fail and will fail. Instead of implementing disaster recovery as an after-thought, multi-site geographic failover must be designed into the application.”
The Citrix report recommends that as companies plan and build their cloud computing models, they think about solutions that allow them to accommodate both traditional application and cloud native workloads. Companies should also plan both short-term and long-term cloud strategies.
As more applications are adopted and use cases expand, it will be even more important for companies to be able to accommodate both types of workloads. Many traditional workloads, for example, are evolving to become more distributed and less reliant on traditional system architecture. Also, establishing separate private clouds for both workload styles would be too inefficient and costly, the report states.
Here is a link to the report: