Design & Build

Data Center Infrastructure, Storage-Design Guide: Network Attached Storage (NAS) Use Cases

by ‎09-26-2013 06:54 AM - edited ‎08-06-2014 08:32 AM (2,605 Views)

Synopsis: Provides design guidance for three use cases where Network Attached Storage (NAS) benefits from a Brocade VCS Fabric and Brocade VDX Switches. The use cases are virtual server storage pools, application pipeline processing and scale-out storage for clouds.

 

 

 

Preface

 

Today’s enterprise storage administrators face many challenges. Between regulation compliance, trends in business analytics, and expanding use of mobile devices, “Big Data” is no longer relegated to supercomputing environments and large institutions. Storage admins must constantly be looking out for new ways to expand their infrastructure to accommodate new data, retain old data longer, and meet the performance needs of new analytics and media applications. On top of capacity and performance requirements, there is increasing pressure to manage costs, and not just capital outlays, but power, cooling, and footprint that part of the total cost of ownership.

 

Traditional scalable, high capacity and high performance storage systems were built on SANs with separate networks designed to accommodate storage-specific data flows. The costs to support dedicated storage networks was justifiable to guarantee reliable performance, especially compared with the inefficient protocols, high latency, and low throughput associated with broadly affordable technologies like Ethernet. But today, more bandwidth and lower latency are available for file storage using modern NAS storage arrays. And, a new NAS architecture, Scale-out NAS, is well suited for applications that rely on scale-out of compute nodes for rapid growth as found in web 2.0 applications hosted by Google, Amazon, Facebook and regional infrastructure as a service (IaaS) hosting companies.

 

While NAS is not a new concept, treating a high-performance NAS solution like legacy file server storage is unlikely to satisfy contemporary business needs. A holistic understanding of the application, data flow, topology, network infrastructure, and file system are vital to designing a successful NAS architecture. This document approaches three different use cases for NAS storage:

  • Virtual Server Storage Pools
  • Application Pipeline Processing
  • Scale-out Storage for Cloud Services

 

Each of these use cases have network requirements that are specialized and beyond the traditional use of NAS as file serving to desktop and business application servers. Each has embraced 10 GbE, manages large data volumes and/or large file sizes and has high availability and data protection requirements as well as dynamic scaling requirements (little to no outage windows).

 

The designs shown here rely on Brocade® VCS® Fabric technology to deliver the simplicity, dynamic scalability and high-performance requirements of NAS storage and the applications and clients that access it.

 

Purpose of This Document

Provides network designs for three different NAS use cases showing how to apply a Brocade VCS Fabric with Brocade VDX switches to meet their unique requirements. A VCS Fabric connects NAS servers, scale-out NAS clusters and client file I/O traffic to NAS storage.

 

Audience

Storage engineers and architects and those focused on cloud computing “at scale” who need high-performance NAS storage with dynamic scalability.

 

Related Documents

 

References

 

About Brocade

Brocade®(NASDAQ: BRCD) networking solutions help the world’s leading organizations transition smoothly to a world where applications and information reside anywhere. This vision is designed to deliver key business benefits such as unmatched simplicity, non-stop networking, application optimization, and investment protection.

 

Innovative Ethernet and storage networking solutions for data center, campus, and service provider networks help reduce complexity and cost while enabling virtualization and cloud computing to increase business agility.

 

To help ensure a complete solution, Brocade partners with world-class IT companies and provides comprehensive education, support, and professional services offerings. (www.brocade.com)

 

Key Contributors

The content in this guide was provided by the following key contributors.

                Lead Architect: Marcus Thordal, Strategic Solutions Lab

 

Document History

Date                       Version                 Description

2013-09-26             1.0                          Initial Version

 

Reference Architecture

The Data Center reference architecture has a NAS template with VCS Fabric leaf-spine building blocks. The template provides connectivity to servers via leaf switches and NAS storage arrays via spine switches. The leaf switches connect to clients which can be on physical or virtual servers.  The template also supports an optional network services block for firewall, intrusion detection/intrusion prevention (IDS/IPS) and server load balancing.

 

Storage-DesginGuide_NASUseCase-ReferenceArchitecture.jpg

    Reference Architecture for NAS Storage Use Cases

 

The VCS Fabric leaf-spine for NAS template provides excellent scalability for both north-south and east-west traffic flows. It provides 10 GbE edge ports and up to 40 GbE Fabric ISL ports. Brocade ISL Trunks logically group multiple inter-switch links (ISL) into a logical trunk. A Brocade ISL Trunk has frame-based, not flow-based, load balancing. This ensures full utilization of all links in the trunk avoiding “hot-spots” commonly found with traditional link aggregation group (LAG) trunks that rely on hashing to place an entire flow on a single link. With a VCS Fabric Trunk, all flows are stripped across all links in the trunk. Changing LAG hashing to move a flow off an over-used link is complicated, time consuming and inefficient requiring an outage window for network maintenance. This is never required with Brocade ISL Trunks as all flows use all links and adding a link to increase trunk bandwidth only requires adding a cable between switches. The rest is automatically taken care of in the VCS Fabric.

 

References

 

Virtual Server with NAS Storage Pool Use Case

VMware supports NAS storage pools and for many environments, NAS storage is preferred. VMware has published best practices and design guidance for deploying vSphere with NAS storage. As more and more application workloads are deployed on ESX servers in a VMware vSphere cluster, these environments need low latency, assurance that file I/O to NAS storage will be available and high availability as loss of connectivity to storage from an ESX server impacts multiple application workloads, not just one.

 

Storage-DesginGuide_NASUseCase-VirtualServerStoragePool.jpg

 

   Virtual Server with NAS Storage Pool Use Case

 

References

 

Application Pipeline Processing Use Case

This category is frequently found in high-performance computing (HPC) environments, graphics processing in the entertainment industry, Big Data analytics, geophysical processing (oil & gas exploration) and signal processing in the defense intelligence community. The amount of storage can be quite large (petabytes), file sizes can be in the Terabyte range and very low latency for file I/O is a critical factor so application processing can complete in every shortening time frames.

Storage-DesginGuide_NASUseCase-ApplicationPipeline.jpg

 

   Typical Application Pipeline Architecture

 

In oil and gas exploration, the advanced technology involved in the exploration and production of oil and gas monitors, analyzes, retrieves, or processes large amounts of file data. Processing needed to analyze it in must happen in real-time.

 

Seismic surveys can produce huge amounts of data that must be captured in ultra-high 3D resolutions and quickly processed for display on computer workstations whose performance rivals mainframe computers. Mulitple renderings of the same large data set are often being maintained simultaneously, requiring multi-petabytes of available data storage space.

 

Similar concerns affect the entertainment industry where computer animation and visual effects, audio processing are becoming common even in movies with live actors. Profits are squeezed if production schedules are delayed, so there is enormous pressure to reduce processing time. In this industry, video editing relies on very high performance workstations that ingest multi-terabyte files and save versions as the editing process progresses. Dynamic scaling of the entire infrastructure (compute, storage and network) with high bandwidth connections (10 GbE, 40 GbE) are essential.

 

Scale-out NAS Use Case

Traditional NAS arrays used larger servers (processing) and large disks to scale-up. But, for applications that grow at web 2.0 rates as found in social media sites including Google, Facebook, Amazon, Zynga, etc., scale-up is not effective. Instead, scale-out architectures are better.

 

Storage-DesginGuide_NASUseCase-ScaleOutNASCluster.jpg

 

   Scale-out NAS Design with VCS Fabric Leaf-Spine Blocks

 

Scale-out NAS provides very high aggregate file I/O since each additional node adds more network ports and bandwidth. However, individual file I/O rates to a node are limited by the number of network connections and the bandwidth of the nodes connection. Using vLAG with Brocade ISL Trunks ensures dynamically scalable bandwidth to every NAS node and full utilization of the Brocade ISL Trunks.  Depending on the network architecture, scale-out NAS nodes can be co-resident with servers in the same rack or housed in separate racks. East-west traffic flows can require a lot of bandwidth and performance demands very low NAS node-to-NAS node latency and client-to-NAS node latency. The Brocade VCS Fabric provides excellent bandwidth for east-west traffic flows at low latency.

 

References

 

Business Requirements

A challenge for data center architects and designers is how to meet the mixed performance demands of a diverse set of business applications. Disparate workloads such as business analytics with large data sets, aka, “Big Data”, back office applications with high rates of transaction performance (e.g., CRM, ERP, financial OLTP systems), create a wide range of network performance requirements. And, specialized applications using pipeline processing in the video entertainment, oil & gas exploration and design and manufacturing sectors create unique network traffic requirements for NAS storage.

 

As data is growing, dynamic scalability is increasingly used to ensure the storage infrastructure keeps up with the business applications ingestion of more and more content. This has driven NAS to scale dynamically. In turn, this had forced networks to also scale dynamically. But, networks have been designed with static scalability requiring administrator intervention and scheduled downtime to reconfigure settings, policies and add more inter-switch bandwidth. This scaling model is deficient. It is being replaced with Ethernet fabrics, software defined networking (SDN) and on-demand orchestration tools including OpenStack. Brocade’s implementation of Ethernet fabrics is the Brocade VCS Fabric. It supports layer-2 and layer-3 fabrics that can scale dynamically without scheduled downtime. For example, VCS Fabric ISL Trunks are dynamically scalable by sampling adding a cable between switches.  Traffic load balancing is at the frame level, and automatically takes advantage of a new 10 GbE or 40 GbE cable with on administrator intervention.

 

Server virtualization clusters running tens of workloads per node and multiple hundreds of workloads per cluster can move any workload across server, storage and network resource pools as resource demands dictate. At scale, this requires a network providing uniform latency, use of all least cost paths, high resiliency and configuration simplicity. Storage and network management have become much more complex in the face of this and data centers commonly include disparate types of storage, storage protocols and network topologies. On top of this, cluster technology, the key to scale-out solutions for compute and storage resources, place much more stringent demands on network bandwidth, latency, scalability and availability often exceeding the capabilities of static hierarchical networks based on classic Ethernet. Network management with repeated entry of commands on multiple switches and ports are required to implement a policy change, adjust bandwidth in a link aggregation group (LAG) or trunk, and to tune around a hot-spot on a physical link in a LAG. The manual management and operation model fails to keep pace with a virtual data center where workload migration changes the network traffic flows, polices have to move when the workload migrates and dynamic storage pools change the latency and bandwidth of storage traffic rapidly.

 

The solution requires a data center network that leverages the same design principles for servers, storage and networking: plug-in resources to scale-out resource pools, automatic dynamic load balancing, and policy defined configuration. Brocade VCS Fabric technology is designed to simplify the network for a wide-range of NAS storage use cases.

 

Special Considerations

Each NAS vendor has specific best practice recommendations and design guidelines for using their products. These should be reviewed and adhered to. In addition, Brocade VCS Fabric technology is constantly advancing.  It is also important to review the Brocade Network Operating Systems (NOS) release notes to ensure correct configuration and support for key features.

 

Design

Each use case has specific design requirements that will influence the configuration of the NAS array as well as the network.  The following provides a set of useful questions to use to help quantify the design choices for the NAS array and the network connectivity.

 

Base Design

The following shows the templates with building blocks used in the base design for NAS storage.

 

Storage-DesginGuide_NASUseCase-ReferenceArchitecture.jpg

…NAS Storage Template

 

Core Template

The Core template provides routing between VCS Fabrics and connectivity to the Internet and Campus network. In some configurations, NAS storage traffic is routed between multiple VCS Fabrics. This adds latency but may provide improved storage utilization at an acceptable increase in file I/O latency.

 

The core routing can use Brocade MLX/ MLXe routers, or existing core routers.  The Data Center Template (see below) uses a VCS Fabric Leaf-Spine with excellent interoperability.

 

For smaller scale environments, the routing can be collapsed into the Data Center Template using Brocade VDX 8770 switches for the Spine Switches reducing cost and complexity.

 

References

 

Data Center Template

The Data Center template uses VCS Fabric building blocks to construct a Leaf-Spine topology. This provides excellent scalability for both east-west and north-south traffic flows. The NAS storage template connects to the Spine block while the servers (physical and virtual) connect to the Leaf block.

 

The diagram below shows the VCS Fabric topology used to validate the design for these use cases.

Storage-DesginGuide_VCSFabricLeafSpineBlock.jpg  

   Brocade VCS Fabric and Network Topology

 

A VCS Fabric connects to multiple NAS server nodes via the Spine and to multiple clients via the Leafs. The Base Design includes a Core Template for core routing with either a Brocade MLX or MLXe router, or existing core routing.

 

An option is to use the VDX 8770 (as shown above) that includes OSPF as well as VRRP-e for excellent scalability and resiliency. And, emerging overlay networking such as VXLAN and VMware NSX software switching are being added to the VDX Switch family.

 

An optional IP Services block provides firewall, application delivery control (load balancing) and IDS/IPS services. This block includes the Brocade ADX Application Delivery switch.

 

The references below include a link to a set of NAS Storage templates well suited for designing the Data Center Template for all three NAS Storage use cases.

 

References

 

Management Template

The Management template provides management, administration, monitoring and configuration services.  It includes Brocade Network Advisor, sFlow and Brocade plug-ins to VMware vCenter if VMware vSphere is used. Brocade is also integrating OpenFlow, OpenStack and OpenDaylight into its products including the Brocade MLX/MLXe Router Family and Brocade VDX Switches.

 

References

 

NAS Storage Template

This template is designed in accordance with the best practices and recommendations of the NAS array vendor. In addition, if NAS is used with virtual servers, such as VMware, Microsoft or Zen hypervisor implementations, then specific configuration and best practices for supporting virtual servers maybe provided.

 

References

 

Server Virtualization Template (Optional)

The template includes the configuration for server virtualization based on the best practices and design guides provided by the particular server virtualization vendor: VMware, Microsoft, Zen, etc. VMware, in particular, publishes specific best practices for NAS storage which should be followed.

 

A description of the Base Design is provided in the references below in the “Storage Infrastructure Design Guide: Scale-out NAS Storage Templates” publication. The Base Design was developed for use with the Scale-Out NAS use case, but is equally effective for Virtualization with NAS and Application Pipeline use cases.

 

References