Server SAN Defined: Architecture and Examples
The next evolution in storage technology is the Server SAN, offering enterprise-level scale and performance. Here's what Server SAN architecture looks like and the benefits it brings to the table.
While there is little doubt amongst IT directors that software-defined storage is having an impact on how storage is managed in the enterprise space, there is still some debate as to what to call this phenomenon. One term, championed by The Wikibon Project, a Marlborough, MA-based open source consulting and research firm, is Server SAN. Read: Server SAN: Demystifying Today's Newest Storage Buzzword
Wikibon describes the Server SAN (storage area network) as a combination of servers and pooled storage connected to multiple servers. Communications between the direct attached storage (DAS) occurs through a high-speed interconnect, such as InfiniBand, low-latency Ethernet, Fibre Channel or another similar high-speed network. Essentially, in a Server SAN, all storage appears to the network to be "local" as if it were attached to the same server backplane, regardless of where in the network the storage physically resides.
When building a traditional SAN, the IT manager first has to decide how big the SAN will be, says Trent Fitz, chief marketing officer at Austin, TX-based Nimboxx Inc., manufacturer of a hyperconverged appliance that combines storage, server, networking and virtualization technology. Once that maximum size is reached, the company either needs to buy an additional SAN or get a larger "box" for the first one. Because Server SANs are self-contained boxes that include the server, storage, networking and virtualization software, expanding a Server SAN is similar to snapping one Lego block on to another.
Because all of the essential technology is included in one appliance, it can scale from serving the needs of a mid-size business to an enterprise. Because the servers are based on open source software, Fitz says, customers do not incur more virtualization software licenses as they grow their network; as they would if they used proprietary software, such as that from VMware or Microsoft.
Server SAN: Defined
Except for some niche markets, the Server SAN is probably more than most small companies require, Fitz says. However, if a company starts out with the Server SAN architecture, it could save upgrade costs down the line as it grows. A traditional data center will include a mix of servers, storage subsystems, network-attached storage and flash or solid-state drives. The data center, he says, can become a complex mixture of infrastructure that grows on an as-needed basis.
Because the Server SAN uses a fully functional appliance that includes all of the requisite components, it can grow in measured steps. Companies that make heavy use of network attached storage (NAS) often find that the devices are "performance challenged," and management is minimal. As users add multiple appliances to their network, Fitz says, they extend the storage pool and scalability, but they are not expanding the capability of the SAN.
|Nimboxx Server SAN Architecture Example|
Nimboxx describes each of its appliances as an atomic unit, Fitz says, because it contains all of the necessary components for a virtual network. Building an appliance-based Server SAN can be compared to adding atoms to each other.
Building a Server SAN also can be compared to building a cloud environment, he says. Some clouds are based on proprietary hypervisors, such as those from VMware or Microsoft. Other clouds are built on open source technology, such as KVM (for Kernel-based Virtual Machine) or the open source Xen hypervisor from Citrix. KVM is part of Linux and not an external hypervisor, so it uses the regular Linux scheduler and memory management. Xen is an external hypervisor that assumes control of the server and divides resources among the guest virtual machines.
Like Nimboxx, EMC also offers a hardware and software converged cloud infrastructure that essentially could be a Server SAN offering. Unlike Nimboxx, EMC, an established storage provider, is partnering with Cisco for the networking expertise and VMware to provide the virtualization software component.
Server SAN Examples
"Server SAN has a large number of potential benefits to application design, application operation, application performance and infrastructure cost. These come from an increased flexibility in how storage is mapped to the applications," according to the Wikibon Project. Server SANs can be used to build specialized sub-networks that would be difficult or impossible to reproduce in traditional array, writes David Floyer in a report on the technology called The Rise of Server SAN.
Among the examples of IT applications are high-performance database subnets that would use remote direct memory access (RDMA) over an InfiniBand point-to-point network, or perhaps a metadata subnet that could process and hold metadata on high fidelity security video (more than 500 million files/day/camera).
Other examples would be a big data subnet that would have knowledge of where data is stored and the data retrieval latencies to optimize big data applications. Such a network could help network optimizer applications to decide when data should be brought to the application on the server and when the application should be moved to the server next to the data. Wikibon also says this approach has benefits for more basic environments -- low-cost file storage using low-performance disks with low-performance Ethernet links and minimal storage services.
The key to optimizing speed, Fitz says, is making sure the data is stored closest to where it will be needed. For example, if the Server SAN appliances are spread throughout an enterprise's corporate facility then the data required by the sales department for its customer database could be stored physically closer to the sales team. Reducing the amount of time data traverses the network can reduce overall network traffic and improve performance.
A rule of thumb, he says, is that all virtual machines (VMs) see the same storage. "Use algorithms to keep the data local to the VM," Fitz says, "then each time the data is accessed it's local."
Ultimately, Wikibon expects the convergence of hyperscale Server SAN and enterprise Server SAN will allow storage to scale more easily and increase the software choices available to infrastructure architects. The company sees Server SANs becoming cost-competitive with hyperscale cloud providers such as Amazon Web Services.
Software-led storage will integrate into software-led infrastructure initiatives such as the orchestration layer of OpenStack, Wikibon says. This will lead to storage management being independent from hardware, allowing greater choice and lower software costs. Open hardware stacks such as the Open Compute Project, together with software stacks like OpenStack, allow enterprise data centers to be cost-competitive with hyperscale vendors and cloud service providers such as AWS.
Will Server SAN Replace Traditional Storage Architecture?
IBM is often credited with determining that 90 percent of the world's data has been created in the past two years. That statistic, however, has been floating around the Internet for at least two years and current estimates are not available. That said, Fitz notes that the traditional storage architecture had been working just fine for the past 25 years. However, the creation of data has skyrocketed and that is forcing changes to the storage environment.
"RAID and (traditional) SANs are no longer viable," he says. "The paradigm for the past 25 years has changed." In order to address the dramatic growth of data, Fitz says companies need to rethink the way data is handled in the network. This includes making the data accessible to everyone who needs it and the process of spinning up virtual servers to manage the massive amounts of data created daily.