DirectFLOW

The DirectFLOW client allows high performance compute clusters to access storage directly and in parallel, providing very high performance and scalability. DirectFLOW distributes the system metadata workload throughout the entire PAS system allowing shared file access and blazing speed without the performance bottlenecks typically encountered in traditional NAS systems.

NFS & CIFS

In addition to the DirectFLOW protocol for Linux servers, the PAS family supports scalable random I/O application needs through UNIX NFS and Windows CIFS data access protocols. By supporting these standard file access protocols, Panasas Scale-Out NAS seamlessly integrates into existing infrastructure to accelerate return on investment.

http://www.panasas.com/products/architecture/network-protocols.html http://www.panasas.com/products/software.php PAS User Interface

Part of the

PAS HC’s efficient design provides nearly one PB of raw capacity in a single floor tile-sized footprint, and capacity can be added without disruption to create vast, multi-petabyte storage pools. Utilizing the patented PanFS™ storage operating system, PAS HC complements the entire PAS family of scale-out NAS solutions to create a single pool of storage under a global namespace. This provides customers with the flexibility to support multiple applications and workflows in a single storage system, blazing performance for complex technical applications and massive capacity to store the results for future analysis.

Panasas DirectorBlades act as a cluster of NAS filers capable of serving data for NFS and CIFS clients. All DirectorBlades are "virtualized" into a single virtual filer, appearing to NFS and CIFS clients as a single storage access point. In this scenario, PanFS enables DirectorBlades to serve as a protocol server, translating NFS and CIFS requests into the native ActiveScale DirectFLOW protocol. Customers can scale performance in near linear fashion simply by adding DirectorBlades.

PanFS has the ability to support all three data access protocols – DirectFlow, NFS and CIFS – simultaneously. Multiple clients can access the Panasas storage cluster using the protocol required by that client. This delivers an unprecedented level of flexibility to optimize the utilization of the storage system.

1. The DirectFLOW client makes a request to the DirectorBlade module to access data files.
2. The DirectorBlade module authenticates the request, obtains the object map of all applicable objects across the StorageBlade modules and sends the map to the client
3. With authentication and a virtual map, clients access data on StorageBlade modules directly and in parallel

This concurrency eliminates the bottleneck of traditional, monolithic storage systems and delivers record-setting data throughput. The number of data streams is limited only by the number of StorageBlade modules and the number of clients in the server cluster.

PanFS has the ability to support all three data access protocols – DirectFlow, NFS and CIFS – simultaneously. Multiple clients can access the Panasas storage cluster using the protocol required by that client. This delivers an unprecedented level of flexibility to optimize the utilization of the storage system.

Shelf/Switch

The Panasas System Shelf and integrated Gigabit Ethernet Switch Blade provide an enterprise-class foundation optimized for simple scalability and reliability. Up to eleven Storage Blade modules and Director Blade modules can be mixed and matched within a single shelf. The 16-port Gigabit Ethernet Switch Blade optimizes I/O scalability by providing line-rate switching between the blades and the data center switches.

Panasas Modules contain 11 blade modules. Shelves can be configured with up to DirectorBlade modules (providing metadata services) and StorageBlade modules (providing intelligent data storage).

 

Panasas Storage Blade modules are the hardware foundation of the system. Storage Blade modules contain all application and user data, handling 90% of file system activity. As a result, Panasas Scale-Out NAS delivers near linear scaling of performance with increases in capacity. Storage Blade modules are configured with Director Blade modules to meet the demands of each specific application environment.

Panasas Director Blade modules uniquely orchestrate file system activity while leaving the data path free to optimize application throughput. This design helps speed data transfers and facilitates better scalability. In addition, Director Blade modules virtualize data objects across all available Storage Blade modules enabling the system to be viewed as a single, easily managed namespace. Director Blade modules will automatically cluster together to support highly scalable storage networks.

PanFS delivers high levels of performance by providing a single namespace and single mount point that reduces administrative complexity and delivers excellent scalability. System reliability and availability are protected by PanFS through Tiered Parity data protection and automatic load balancing. PanFS supports DirectFlow (pNFS), NFS and CIFS data access protocols simultaneously, delivering an unprecedented level of flexibility and utilization.

In fact, the idea for pNFS was born in 2003 out of conversations between Garth Gibson (CMU and Founder and CTO at Panasas,) Gary Grider of LANL and Lee Ward of Sandia. This initial brainstorming was followed up with a Workshop on NFS Extensions for Parallel Storage in late 2003 chaired by Gibson and Peter Honeyman of the Center for Information Technology Integration at the University of Michigan.

More collaboration and conversation led to the submission of the original problem statement to the IETF in 2004. This was authored by Gibson and Brent Welch of Panasas along with Peter Corbett of Network Appliance. From this point on industry support and involvement steadily increased and led to the IETF working group folding pNFS into the NFSv4.1 minor revision draft in 2006. Gibson, Welch and many others from Panasas continue to be heavily involved in driving this standard forward.

• Just getting started? Then take a look at the Parallel Primer and you’ll get the basics down in no time
• Nothing but code? You need Parallel Particulars. It's all here (specs, projects, code…)

For the last few years, high-performance data centers have been moving aggressively towards parallel technologies, such as clustered computing and multi-core processors, which accelerated development and deployment of parallel applications.

While this increased use of parallelism solves the vast majority of computational bottlenecks, it shifted the performance bottlenecks to the storage I/O system. With mainstream computing going parallel, the storage subsystem needs to migrate to parallel technology. To become ubiquitous, a standard approach which allows choices from multiple storage vendors and the freedom to access parallel storage from any client is required.

To move to the next level of performance, storage systems must be optimized for parallelism, while adhering to an economically efficient standard. NFS, the current network file system standard, doesn’t support parallel I/O and existing parallel products from the key storage vendors are not compatible with each other. Until the industry delivers a parallel storage standard, user adoption will continue to be hampered by their reluctance to deploy one of the many incompatible parallel storage implementations.

Later this year, the Internet Engineering Task Force (IETF) NFSv4 subcommittee is expected to conclude its work on the Parallel NFS (pNFS) protocol that is part of the NFS version 4.1 RFC. This milestone will move NFS version 4.1 from Internet-Draft to a proposed standard. Parallel NFS enables direct parallel data transfer between clients and storage devices without the need for expensive filer heads. Support is expected for Linux, Windows, and the leading UNIX versions, such as Solaris and AIX.

This new standard is being developed by a consortium of storage industry technology leaders, including Panasas, IBM, EMC, Network Appliance, Sun Microsystems, and University of Michigan’s Center for Information Technology Integration (CITI).

In order to understand how pNFS works it is first necessary to understand what goes on in a typical NFS architecture when a client attempts to access a file. Figure 1 shows a traditional NFS architecture. You can see that the NFS server sits between the client computer and the actual physical storage devices. When the client wants to access files residing on that storage it must create a link to the NFS server (known as creating a mount point.) When the client attempts to access files the NFS server acts as an intermediary and manages all of the data processing required to deliver data back to the client requesting it.

This architecture works well for relatively small data sets being accessed by a few clients and provides significant benefits over Direct Attached Storage (like the disk in your pc); namely that data can be shared by multiple clients and accessed by any client that has NFS capabilities. However if large numbers of clients need access to the data or the data set grows too large then the NFS server quickly becomes a bottleneck and chokes the system performance. Fundamentally pNFS removes that bottleneck allowing incredibly fast access to very large data sets from many many clients.

Here we see how pNFS modifies the NFS architecture to eliminate the bottleneck we just described. Essentially the NFS server is moved ‘out-of-band’ and becomes what is known as a metadata server. That means it manages data about data. So when a client needs to access data what does it do?

The first thing it does is talk to the NFS server just as it did in the previous example. However this time the server provides the client with a map of where to find the data and credentials regarding its rights to read/modify/write the data. Once the client has those two components, it talks directly to the storage devices when accessing the data. With traditional NFS every bit of data flows through the NFS server – with pNFS the NFS server is removed from the primary data path allowing free and fast access to data. Of course all the advantages of NFS are maintained but now there is no bottleneck and data can be accessed in parallel allowing for very fast throughput rates and system capacity can be easily scaled without impacting overall performance.

• PAS 7 - Full-Featured, Entry System with Balanced Performance and Scalability providing up to 44 TB of raw capacity in a compact 4U rack-unit (7”) shelf.
• PAS 8 - Excellent Performance and Scalability for highly sequential I/O workloads and large files supporting a broad range of scientific and technical applications
• PAS 9 - Blazing Performance for random I/O workloads and large and small files. PAS 9 utilizes solid state disks (SSD) that accelerates metadata operations providing extreme levels of throughput for performance-sensitive applications