Dell Storage PS Series Architecture: Load Balancers Dell EMC Engineering February 2017 A Dell EMC Technical White Paper
Revisions Date Description August 2011 Initial release February 2017 Updated to reflect new branding and formatting Acknowledgements This paper was produced by the following members of the Dell EMC™ Storage team: Engineering: Chuck Armstrong The information in this publication is provided “as is.” Dell Inc.
Table of contents 1 Introduction ...................................................................................................................................................................5 2 PS Series overview ......................................................................................................................................................6 3 Why automatically balance SAN resources ..............................................................................................
Executive summary This document provides VMware® and Dell™ PS Series storage administrators with recommended best practices when configuring a virtual environment. It describes the function of PS Series load balancers, which are used to intelligently maintain system performance at optimum levels with PS Series storage. This document also covers connectivity and high availability, as well as some common performance-tuning best practices.
1 Introduction In many organizations, significant effort is devoted to tuning resources: modifying network settings on switches, changing various parameters on servers, re-configuring QoS on the WAN, and continually altering the layout of data on the SAN. Estimates indicate that 80 percent of most IT budgets is dedicated to operations and maintenance, leaving only 20 percent for technical innovation to enhance the business.
2 PS Series overview Dell PS Series storage is an award-winning, virtualized, scale-out SAN architecture that provides simplified management, ease of use, a comprehensive set of data protection and recovery tools, and great ROI for both large and small organizations. PS Series arrays are designed and built to be highly available and work together to provide virtualized SAN resources called a Group.
3 Why automatically balance SAN resources The IT resources available to most organizations are limited. The realities of budgets, staffing, knowledge, and the business needs of the organization are often competing and conflicting, forcing IT managers to optimize the environment for the proper balance of these items. The usual result is that tuning for optimal performance is deprioritized in favor of more pressing items.
4 PS Series load balancers in PS Series storage pools When you initialize the first array and create a PS Series Group, a default pool is automatically established. An array added to the Group is referred to as a member of the Group. The first member added to the Group is initially placed into the default pool, and administrators subsequently deploy volumes from this pool. Within the pool, resources such as network bandwidth, disk capacity, and I/O are balanced automatically.
5 How the Network Load Balancer (NLB) works Communications between application servers (iSCSI initiators) and volumes (iSCSI targets) are called connections. A PS Series Group will present all iSCSI targets through a single virtual address known as the Group IP address. This allows administrators to establish connections easily by configuring the iSCSI initiator with only the Group IP address instead of the IP addresses of all of the interfaces in the Group.
6 How the Capacity Load Balancer (CLB) works The CLB ensures that as volumes are created and deleted, and as members are added to and removed from a pool, the relative percent of capacity in use is maintained at a consistent level among the members in that pool. Keeping the members in the pool filled to the same percentage of their disk capacity helps to ensure that all of the resources in the pool are used equally, and helps avoid overloading one member compared to another.
other members in the pool. An FSTP will cancel other RBPs. Once the low space issue that prompted the FSTP has been resolved, the CLB will create new RBPs if they are required. 2 All data movement, regardless of whether caused by an RBP or FSTP, is handled in a transactional manner — data is only removed from the source of the transfer and internal metadata that tracks the location of the data is updated only after its receipt is confirmed by the target of the transfer.
7 How the Automatic Performance Load Balancer (APLB) works The APLB feature is designed to help alleviate the difficulties inherent in manually balancing the utilization of SAN performance resources. Operating on the resources in a pool, The APLB is capable of adjusting to dynamic workloads in real time and at a sub-volume level.
Example of APLB in a non-tiered environment The APLB is surprisingly simple in its concept and execution, leveraging various aspects of the PS Series architecture to automatically balance the performance delivered to applications by the PS Series SAN. For example, the rebalance plans that the CLB uses to re-adjust the placement of data, are leveraged by the APLB as well.
which is a composite score of factors such as RAID type, disk speed, number of disks, as well as PS Series controller type and the current I/O load. The member with the lower relative busy-ness will be chosen for the data exchange. The APLB works well in a variety of environments. For example, in PS Series pools with members displaying similar performance characteristics, the net effect is to eliminate hot spots in the pool. This is shown in Figure 1.
Reduction of worst-case-scenario purchasing: By working continually, the APLB can detect and act on cyclical business processes, such as increased end-of-month activity by the finance group enabling the resources of the SAN to be leveraged in near-real-time. This may enable IT management to purchase fewer resources since each application can better leverage the storage when it needs it most.
8 Observing latencies and high I/O data Various metrics central to the APLB behavior (such as array latencies and breakdown of data classified as high, medium, and low I/O) are displayed using SAN HeadQuarters. These metrics are part of the I/O subsection of each Group and can be readily observed for Groups, pool and the individual members within a pool. Observing this information can be helpful in understanding the operation of the APLB.
9 Tiering with the APLB When provided with tiered resources in a pool, such as arrays with different spindle speeds or set to different RAID types, the APLB is able to use them to tier the workload. This is not limited to any particular RAID type, interface I/O type, spindle speed, number of disks, or PS Series controller generation, since all of these factors are abstracted by the use of latency, which is the primary factor when deciding when to rebalance the workload.
10 Using the APLB with hybrid arrays The PS Series hybrid arrays (such as the PS4210XS, PS6210XS, and PS6610ES) use their own internal implementation of the APLB to compensate for unbalanced or tiered workloads. These arrays shift the active parts of the workload in the volume slices that they are responsible for to SSD media for extremely lowlatency I/O, while leaving the bulk of the workload on HDD media for normal latency I/O.
Example of APLB in a tiered environment 19 Dell Storage PS Series Architecture: Load Balancers | TR1070
11 Scenarios in which APLB action is limited There are several cases when the APLB will not take action. First, if all of the members of the storage pool are exhibiting low latencies, then the resources do not need to be balanced.
Fourth, if a customer has chosen to override the automatic placement of volumes by the capacity balancer by using the CLI bind command, all balancing of any sort is disabled for that volume. Note that this does not disable the APLB for other application workloads that may be using the same member that the volume is bound to. A case where the APLB behavior should be observed more carefully is when volume RAID preference is utilized.
12 APLB and backup operations Backup operations place a large burden on storage resources but should not be a problem for the APLB. Backup operations largely result in sequential reads, and only access a given portion of the data for a brief portion of the function. Thus, since there is no repeating pattern of access, the APLB will not take action, which is desirable since there is no optimization that may be achieved.
13 Disabling the APLB While not recommended, the APLB may be disabled by unchecking the Enable Performance Balancing in Pools checkbox in the GUI. Note that this is a Group-wide setting and will affect all pools in the PS Series Group.
14 Summary The various load balancers that work in a PS Series pool provide a very flexible, dynamic framework that is able to quickly adapt the PS Series storage to the shifting requirements of the customer environment. Changing I/O activity is easily balanced between the available network interfaces on an iSCSI-connectionbasis by the NLB. Working in conjunction, the CLB and the APLB are able to help actively manage the data on the members using whatever resources the customer has chosen to provide.
A Additional examples of APLB The following are some additional examples of the data distribution before and after APLB has performed its load balancing function.
Rebalancing after adding performance-optimized resources to a pool 26 Dell Storage PS Series Architecture: Load Balancers | TR1070
Rebalancing with multiple hybrid arrays 27 Dell Storage PS Series Architecture: Load Balancers | TR1070
B Technical support and resources Dell.com/Support is focused on meeting customer needs with proven services and support. Dell TechCenter is an online technical community where IT professionals have access to numerous resources for Dell software, hardware and services. Storage Solutions Technical Documents on Dell TechCenter provide expertise that helps to ensure customer success on Dell EMC storage platforms. B.