Use Storage QoS to Manage Performance for Multiple Workloads
In highly virtualized and cloud environments,storage quality of service (QoS) is becoming more and more critical for preventing workloads or tenants from adversely affecting one another and for meeting service-level objectives for storage performance. The need for storage QoS in virtualized data centers is being driven by the accelerated use of shared storage infrastructures to support multiple workloads. Those organizations deploying public and private clouds have gone further, adding automation, policy-driven provisioning and multi-tenant architectures to support service-level agreements (SLAs) for a diverse group of users, all of whom may be running on a common infrastructure. Let’s look at three use cases where storage QoS can make a critical difference in managing performance for multiple workloads.
Gain in-depth insights into NetApp’s clustered Data ONTAP® by joining an upcoming webcast.
Storage QoS Example 1: Throttle Performance for Misbehaving Applications
Perhaps the most common use case for managing storage QoS is the need to throttle an application that is hogging resources at the expense of other workloads. Identifying the rogue application, notifying the owner and correcting the application can be a time-consuming process. For most storage administrators, a simpler and faster solution is to throttle the resources for the offending LUN or volume. Advanced storage QoS features enable a storage administrator to throttle throughput in terms of IOPs or MB/second for multiple data container types, such as a storage virtual machine (SVM), volumes, LUNS, or VMDK files within an SVM.
Storage QoS Example 2: Set Storage Performance Limits for Specific Workloads
One approach to managing QoS for workloads with different service-level needs, such as development and test, is to house them in dedicated infrastructure silos. However, this leads to lower resource utilization and higher costs. With storage QoS, it is possible to run both types of workloads on a cost-effective, shared infrastructure by assigning the data containers mapping to the dev/test workloads to a policy group with a throughput limit so that they do not adversely affect the production workloads. With NetApp® Data ONTAP®, you can define up to 3,500 different policy groups in a storage cluster.
Storage QoS Example 3: Limit Maximum Storage Performance for Cloud Tenants
Cloud providers, whether internal or external, can be faced with a different type of QoS problem. In a multi-tenant cloud environment, the first customers may enjoy access to infrastructure resources over and above their service-level agreements (SLAs). This can be great while it lasts, but can also create a perception of performance degradation as additional customers are added to the shared storage infrastructure, and those early customers are required to operate within their SLA. In multi-tenant environments, it is often desirable to set a maximum performance limit for each tenant (per their SLA) in order to manage performance expectations as additional customers are onboarded and overall resource utilization increases.
NetApp Data ONTAP and Storage QoS
Data ONTAP 8.2 introduces new QoS capabilities to simplify performance management for a shared infrastructure environment, enabling administrators to quickly and easily allocate shared storage resources to applications, servers, departments or cloud tenants. For more information, or to engage with NetApp experts and discuss best-practice recommendations for Storage QoS, visit the NetApp community page for Data ONTAP.
© 2014 NetApp, Inc. All rights reserved.
No portions of this document may be reproduced without prior written consent of NetApp, Inc. Specifications are subject to change without notice. NetApp, the NetApp logo and Go further, faster, are trademarks or registered trademarks of NetApp, Inc. in the United States and/or other countries. Click here for a full listing of NetApp trademarks. All other brands or products are trademarks or registered trademarks of their respective holders and should be treated as such.