What do virtualization executives think about 2009? A VMBlog.com Series Exclusive.
Contributed by Clive Cook, CEO of RNA networks
Memory Virtualization, the Third Wave of Virtualization
Memory Virtualization, like server and storage virtualization, offers the benefits of consolidation and compelling cost savings. Yet over and above early server and storage implementations, in 2009 Memory Virtualization will introduce a new way of thinking about virtualization that challenges the way IT manages applications, clustering, the data center, services and business itself.
Introduction to Memory Virtualization
In simple terms, virtualization is the abstraction of IT resources, separating their physical instance and boundaries from their function. Virtualization has brought important innovation to IT. Application virtualization started two decades ago followed by server and desktop initiatives. However, server and storage virtualization are but a starting point for the power yet to come from virtualization. The next wave of virtualization changes everything.
The greatest benefit these early implementations of virtualization deliver is consolidation – fewer servers, one mainframe to run multiple apps, higher capacity networked storage - resulting in tremendous cost savings. Memory Virtualization delivers this same benefit and more, changing the economics of IT.
What is Memory Virtualization?’ Let’s begin by looking at the role of memory in IT.
Memory is required in every digital machine; switches, routers, appliances and servers. Each contains physical memory alongside the logic that manipulates the 1’s and 0’s. Memory is closely coupled with compute logic, and when performance gains are needed enterprises typically add more memory, which can be very expensive.
For the first time, Memory Virtualization introduces a way to decouple memory from the processor, AND, from the server to provide a shared, distributed or networked function. This is not more addressable memory but virtualized memory shared between multiple machines.
When we talk about Memory Virtualization, it is important to keep in mind that memory and storage are not synonymous. Memory Virtualization is focused on application performance and has a direct touch point with end users. The CPU actively and directly uses data from memory. On the other hand, storage is the persistent store of data. Storage is static, regardless of whether it is a spinning disk, SSD or RAM disk. Data is retrieved from disk and put into memory BEFORE the processor can transform data to information. Adding storage doesn’t solve memory problems, and doesn’t noticeably accelerate application performance.
Now that we know what it is, let’s look at where Memory Virtualization can be immediately applied across IT infrastructures.
- Extending memory beyond a physical server’s capacity.
- Implementing shared memory for clustered or grid computing environments.
- Enabling Cloud Computing and Real-Time Infrastructure (RTI) in the enterprise data center.
In 2009, we will see Memory Virtualization implemented for these Use Cases; Extending Memory, Implementing Shared Memory, and Enabling Cloud Computing and the Next Generation Data Center. The challenge facing Memory Virtualization is to overcome the many and sometimes unrecognized workarounds to not having access to sufficient memory.
Use Case #1: Extending Memory
An application’s working data set is frequently larger than the available physical memory in the server. Today’s single server memory capacity ranges from 1 GB to greater than 64 GB. Memory is a captive resource to the CPU its connected to, yet the working data set of many applications is well beyond this.
Using Memory Virtualization, the entire working data set can be loaded into memory for the processor to access directly, without going to disk.
The benefits to the end user and IT organization of extending memory in this scenario are tremendous. Memory Virtualization implementations consistently show 10-30X performance improvement to the end user. Direct cost savings through server consolidation, reduced over-provisioning, deferring server upgrades, and increased utilization are immediately realized with Memory Virtualization.
Memory Virtualization paired with x86 Server Virtualization will drive the adoption of virtual machines (VM’s) into the production data center. Today, Server Virtualization has been limited to operational applications, not business critical applications due to performance and reliability concerns.
Use Case #2: Implementing Shared Memory
Agility to cost-effectively deliver new service levels is a key use case for Memory Virtualization. By decoupling physical resources and distributing them across a network, IT services can be provisioned on-demand with a usage based payment model. With Memory Virtualization, services are no longer tied to resources. Plus, the user and IT organization gain all of the performance and cost benefits of extending memory through Memory Virtualization.
In the shared resource model, applications or services use only as much memory as required. Memory is distributed and shared as an available network resource. Multiple terabytes of memory can be readily available to any application.
In this scenario, companies will now be able to harness vast amounts of available memory in their expensive high-performance servers. Higher end machines often have 2X or more invested in memory than in the CPU or storage. With Memory Virtualization ROI is immediate and significant.
Use Case #3: Enabling Cloud Computing and other Next Generation Data Center Initiatives
The choice to use a next generation data center or a Cloud Provider is an important strategic decision for any enterprise. Memory Virtualization plays an enabling role in both environments. Memory Virtualization is critical to implementing agile services for these infrastructures.
By offering memory as a pooled resource in the data center, Memory Virtualization delivers dramatic cost savings from reduced power and cooling. Real-time Infrastructure is available at 30% less cost in a pooled resource architecture versus over a tiered architecture.
Memory Virtualization is an enabler of Cloud Computing in this use case. Memory Virtualization enables different infrastructure Cloud Models to be available as different services with fungible resources. Memory resources underpin the MetaOS implementations (i.e. Microsoft Azure, Amazon EC2) that will govern these Clouds. Virtual memory provides the OS and application memory for this distributed operating system.
Conclusion: Memory Virtualization is a Big Idea Who’s Time has Come
To conclude, Memory Virtualization can dramatically boost performance at significantly lower cost while delivering service levels that will transform business. Making memory a truly shared, network resource has broad and deep implications across the spectrum of enterprise applications, clustered computing and data center operations. IT architects who are planning both cloud and corporate data center transformations in 2009 have much to gain with Memory Virtualization, given limitations of the current memory model.
About Clive Cook
Clive Cook is CEO of RNA networks http://www.rnanetworks.com/, a pioneer of memory virtualization that dramatically advances data center utilization and performance.