Posted by nevex on May 8, 2012
At first glance, the software world appears to be dominated by Microsoft Windows. However, Linux is more prevalent across a number of platforms than most people realize. It’s the dominant OS for everything from supercomputers down to smart phones and tablets that run the Linux-based Android OS.
Microsoft Widows is still the dominant player based on raw numbers but looking at the bigger picture, the momentum is behind Linux for enterprise server applications. Why? Because large corporations are concerned about cost and security.
We see the OS market divided into three categories:
- Windows for client machines and laptops
- Windows Servers for SMB users
- Linux and Windows Servers for Enterprise users
A recent survey by the Linux Foundation “Linux Adoption Trends 2012: A Survey of Enterprise End Users” showed the growth is coming from cloud computing, big data and “greenfield” deployments as new opportunities, complementing the current dominance of Linux in web server applications. The survey made it clear that enterprise users are flocking to Linux servers for the TCO, features set and overall security benefits.
Significant growth in Linux is not simply expansion at incumbent customer sites. More than 70 percent of new Linux deployments during the past two years have been for new applications and greenfield installations, nearly twice that number were migrations from the Windows server.
Where are all of these new Linux installations being used? The survey revealed two key Linux growth markets: big data and cloud computing. More than three-quarters of the respondents expressed concerns about handling the growth of big data and 72 percent expected to use Linux servers in their big data plans.
The survey also showed strong worldwide growth for cloud computing with 61 percent of the organizations now using cloud-based applications and two-thirds using Linux as their primary cloud OS. Cloud computing platforms depend on a highly virtualized infrastructure to derive efficiency, productivity and flexibility. Given the large deployments of cloud computing, it’s not surprising that 72 percent of organizations have more than one-quarter of their servers virtualized, while 46 percent expect to have up to three-quarters of their servers virtual by the end of this year.
This massive shift towards virtualized computing holds the potential for sunny cost savings but it also holds potential for a mighty downpour. As enterprises plan new Linux deployments across private, public or hybrid clouds, those with virtualized environments will need to pay special attention to the storage I/O bottleneck, endemic in virtual server installations.
Server virtualization boosts the resource utilization of physical servers, taking advantage of the power of multicore CPUs to share across multiple virtual machines. But storage I/O performance was not built with virtualized environments in mind. Running a single data-intensive application on a fast physical server can choke the I/O to attached hard disk storage, so packing dozens of virtual machines across the same I/O bandwidth is a sure way to kill application performance and create legions of angry users. And that’s where intelligent caching software comes to the rescue.
There’s little doubt a Linux server is in your future and chances are good that it will be virtualized too. Count on us to get you there.
More on that later,
Steve
May 8th, 2012
Posted by nevex on March 8, 2012
A blog by Steven Lamb, CEO
NEVEX Virtual Technologies was conceived as a solutions company, not a pure software play. So when we went looking for a SSD hardware partner, it had to be a company with an SSD hardware platform that was performance-matched to the market-leading speed of our NEVEX CacheWorks software. Texas Memory Systems was an obvious choice: the company is well-established and has the well-earned reputation for consistently delivering the World’s Fastest Storage. The power behind TMS market leadership is its engineering superiority backed by an intense understanding of how SSDs can be optimized to meet the needs of specific applications.
CacheWorks software, with its ability to target acceleration to specific applications rather than accelerating everything in the cache, and its multi-level caching technology that leverages a combination of memory, SSD and DRAM, drove the Texas Memory Systems’ RamSan-70 PCIe flash card to unprecedented performance standards. We’ll soon be publishing performance benchmarks for the new NEVEX-TMS solution but it’s safe to say this is truly a case of the world’s fastest cache powering the world’s fastest flash.
The NEVEX-TMS partnership resulted from the recognition that for flash to gain deeper penetration into the enterprise market, intelligent software was required. Our state-of-the-art caching software opens up the server-side caching market for flash vendors, such as TMS. Those in the know have been quick to recognize this hot and rapidly growing market. Witness the recent acquisition activities of hardware vendors such as Fusion IO-IO Turbine, SanDisk-FlashSoft and OCZ-Sanrad.
Now that the industry has begun paying close attention to server-side caching, we believe the competitive landscape is about to ratchet upwards. The deciding factor will be not who offers the best cache but who is best at delivering white-hot data to the applications that need it. Enterprise customers will be looking for solution providers who can deliver high-performance SLAs utilizing application-specific caching.
Beyond the TMS relationship, we intend to bring our software expertise to application providers who can leverage the Selective Optimized Caching technology within CacheWorks to accelerate any and all enterprise applications that are currently constrained by the I/O-bottleneck. Working with them will expand the market for NEVEX software and SSD hardware vendors will benefit, too.
The potential for growth of server-side caching in the enterprise is enormous. It’s an exciting time for all of us.
Until next time,
Steve
March 8th, 2012
Posted by nevex on December 20, 2011
A blog by Steven Lamb, CEO
Whenever there’s an IOPS/throughput performance mismatch between servers and storage, requiring both server and storage administrators to make ongoing pervasive adjustments, the end result is systems that are held back from reaching their full performance potential. Diagnosing the reasons behind poor performance in today’s complex data infrastructures can be extremely difficult, especially when the issues are intermittent. Much trial and error is often required — a huge time commitment — and the cause of much anxiety in an IT team under the gun to resolve issues quickly.
This is a chronic problem for organizations, especially in today’s increasingly virtualized infrastructure. Traditional “solutions” that attempt to address these issues by fine-tuning the applications or adding ever-increasing amounts of server/memory and storage hardware often fall short in addressing the real root cause of I/O bottlenecks. Using common workarounds such as short stroking HDDs in the storage array, adding SSDs in the storage array or installing PCIe Flash cards, are simply not the silver bullet they purport to be. They are all labor intensive; often complicate the situation further by creating new problems that managers then have to redress, and ultimately contribute to a higher Total Cost of Ownership.
A new approach is available that allows for the acceleration of specific applications; makes better use of existing storage; is transparent to users, applications and storage; works with server virtualization, SAN, DAS, NAS and Cloud storage; and is interoperable with OS file caching. That’s a mighty long list of requirements; NEVEX CacheWorks meets or exceeds all of them as well as being Flash card and storage system agnostic.
Because CacheWorks is policy-based, administrators have the ability to target critical business applications and apply acceleration to them specifically. Whether the applications are running on VMware, Microsoft Hyper-V virtual machines or physical servers, CacheWorks provides industry-unique control over the Windows memory cache to create a multi-level caching solution that can allow applications to exceed the performance of running fully on Flash.
If your organization is facing performance issues and traditional approaches have persistently fallen short, and if you want your systems to reach their full performance potential, consider our new and innovative software that provides application-specific acceleration – and say goodbye to I/O bottlenecks.
Until next time,
Steve
December 20th, 2011
Posted by nevex on December 8, 2011
A blog by Nigel Miller, VP Business Development
Application Performance is the constant headache that IT professionals face every day. Development of affordable Solid State Disk technology has relieved some of the symptoms and opened up a new revenue stream for storage VARs. But it isn’t a complete cure as it only addresses part of the headache. There is a comprehensive cure available, however, and it offers additional profits for VARs: intelligent caching software.
NEVEX CacheWorks is the missing ingredient, the aspirin that enables VARs to offer a complete cure. Bundling our caching software with the VAR’s preferred SSD/Flash hardware addresses the specific customer headache of poor application performance. This combination enables VARs to expand their offering and improve their bottom line.
The appeal for storage VARs to sell CacheWorks via the NEVEX Channel Partnership Program is simple:
•CacheWorks is a drop-in solution that cures the customer’s application performance headache and puts additional profit in the VAR’s pocket;
•CacheWorks runs on 64-bit Windows Servers, physical and virtual machines;
•NEVEX offers a comprehensive channel development program and lucrative margins;
•The breakthrough innovations in CacheWorks provide VARs with the engine to transform the economics of their customers’ storage infrastructures.
NEVEX technology gives VARs the unfair advantage to drive new business and grow revenue. How? NEVEX doesn’t simply accelerate general I/O performance, its caching software goes much further, delivering precise application acceleration where and when it’s needed. CacheWorks is the first caching software that puts customers in control to specify which applications and data are most critical to their business. Unlike conventional caching software that simply promotes the most active data from within a specified LUN and ejects data already resident in the cache, CacheWorks prioritizes targeted applications, ensuring the most business critical applications’ data stays cached for maximum performance. This functionality allows IT administrators to drill down and accelerate the applications that are causing the performance headache. CacheWorks is the aspirin that enables VARs to deliver the cure.
It’s this combination of precisely targeted application acceleration and economic value that makes CacheWorks such a compelling and powerful product for VARs to sell. Until the end of January, NEVEX is offering an Early Adopter Program for VARs that specialize in database optimization, virtualization enterprise application storage and server infrastructure. To learn more about the NEVEX Channel Partnership Program and Early Adopter specials, interested VARs are invited to contact NEVEX Business Development at partners@nevex.com.
Nigel Miller, VP Business Development
December 8th, 2011
Posted by nevex on November 11, 2011
A blog by Rayan Zachariassen, CTO
The long-standing debate about which is better – block-based or file-based storage (or loosely: SAN vs. NAS) – has emerged again to characterize the solutions in our world of SSD caching. Most end-users are fundamentally concerned with economic efficiency. If you weren’t you’d just put your entire environment onto SSD and be done with it.
Economic efficiency has three main aspects to it: performance value, resource efficiency, and management cost. On Monday, I wrote about the first aspect, performance value. Wednesday’s blog focused on the second aspect, resource efficiency. Today’s blog centers on the lower management overhead inherent with file-based caching:
Management Cost
Block caches by their nature are tied to block-based primary storage devices. This means that,
a) They need to be managed and set up explicitly for each storage device, and,
b) They do not work with file-based services such as NAS arrays, or cloud storage gateways, or clustered file systems.
Associating a cache with a particular storage device means that applications remain tied to specific storage devices. This inherently increases the complexity and inter-dependencies, and therefore risk, of the IT infrastructure. One of the big advantages of NAS, for example, is to be able to easily organize data and applications in a flexible namespace independent of the underlying infrastructure, whereas with block-based primary storage you need to maintain the link between application, data, and device. With file-based caching, you now also have the ability to specify the performance you want for any part of your namespace (or application, or project, etc.), which allows administrators to operate at the level of policy decisions as opposed to hardware/software dependencies. Clearly, this makes better use of people time and reduces risk to the organization.
Furthermore, with file-based caching a single technology is able to deal with essentially all caching requirements of an IT infrastructure: DAS, SAN, NAS, Cloud, and futures, independently of how storage devices evolve. This means you can eventually have a single caching solution for your infrastructure instead of a different solution for each kind of storage system you have, which means fewer point solutions, which again reduces the complexity, and therefore cost, of your infrastructure.
And one last thing to consider…Microsoft decided 20+ years ago that file-based kernel buffer caching made for a better solution and built the Windows cache manager on that model, which was contrary to previous block-based buffer caches. Their reasons included more efficient cache processing and improved recoverability. It’s worked pretty well so far.
Rayan Zachariassen, CTO
November 11th, 2011
Posted by nevex on November 9, 2011
A blog by Rayan Zachariassen, CTO
The long-standing debate about which is better – block-based or file-based storage (or loosely: SAN vs. NAS) – has emerged again to characterize the solutions in our world of SSD caching. Most end-users are fundamentally concerned with economic efficiency. If you weren’t you’d just put your entire environment onto SSD and be done with it.
Economic efficiency has three main aspects to it: performance value, resource efficiency, and management cost. On Mon Nov 7, I wrote about performance value. Today’s blog focuses on the second aspect:
Resource Efficiency
File-based caching enables selective optimized caching, where decisions about data to be cached aren’t made simply by which data is the hottest. Instead, administrators can define critical applications and data to be accelerated so that only files associated with specific business needs can be promoted to the cache.
This ability means that the precious cache resource that in a block-based approach would be polluted with any and all data that happens to be active, can now be used exclusively for important data; the data that justified purchasing a cache in the first place. With less contention for the cache resource, a file cache can hold more of the data that matters, which means either better performance (higher cache hit rates) or a lower capacity and therefore less costly cache resource for the same performance.
Either way, the cache efficiency enabled by selective caching based on context provides an advantage that completely sets file-based caching apart from block caching technology.
If you want to dig a bit deeper, here are a couple of relevant reads on this issue: the CTO of Nasuni discussing the block vs. file choice in the context of cloud gateways, and, just to show this is not an entirely new discussion, an article from a decade ago by Solid Data Systems talking about how moving busy files to SSD storage keeps the performance curve going.
Part III of this blog, covering management cost, will be posted on Fri Nov 11.
Rayan Zachariassen, CTO
November 9th, 2011
Posted by nevex on November 7, 2011
A blog by Rayan Zachariassen, CTO
The long-standing debate about which is better – block-based or file-based storage (or loosely: SAN vs. NAS) – has emerged again to characterize the solutions in our world of SSD caching. Here’s my take on it:
Most end-users are fundamentally concerned with economic efficiency. If you weren’t you’d just put your entire environment onto SSD and be done with it.
Economic efficiency has three main aspects to it: performance value, resource efficiency, and management cost, and I’ll go through the major block vs. file issues for each of these in turn.
Blogs, by their very nature, are meant to be fast reads, and as economic efficiency cannot, in my view, be covered off in just 400 words, I have divided it into three segments. So, here are my views on performance value:
Performance Value
Block-based caching is a relatively straightforward technology that essentially maps blocks from primary storage into a block cache. The main advantage of this technology, from a performance perspective, is simplicity – the job is clear, the options limited. There have been heroic attempts in the industry of making block caching intelligent by trying to figure out the relationships between blocks by watching the activity. It’s sort of like looking at a busy highway, trying to figure out which cars are driving in a convoy. You want to know this because there are special lanes for some convoys but they have to be directed to go there. Similarly, if you know the relationship between blocks you can treat related blocks differently (cache them) from the great wash of noisy (car or block) traffic you’re normally looking at.
With file-based caching, no guessing is needed. The relationship between blocks is clear, and in addition you know the context. Imagine watching a highway full of cars and knowing who the driver is in each car, where they came from, why they are on the highway, whether or not they’re in a hurry, and which other cars are going in the same direction. What that means is that with file-based caching one can focus the increased performance where it matters (the special lanes for selected traffic) instead of building a bigger highway for everyone.
The traditional objection to file-based caching is that block caches have to be faster, just because they are so much simpler. In fact, block caches are inherently limited because they have to wait until an I/O request is at the level of a device driver before a block cache can react. File caches operate at a higher level of the operating system where there is access to context, some control over what the OS does with file data, and the ability to react much earlier to an I/O operation. That last point means file caches are the inherently faster technology because they can short-circuit much of the processing that happens before a block cache even sees the I/O.
The NEVEX CacheWorks file-based cache takes advantage of the high level it operates at within the Windows OS to leverage the in-memory buffer cache and incorporate it into caching policies for the SSD cache, which block-based caches simply cannot do on their own. That ability, plus selective performance due to context awareness, plus a shorter I/O path, makes file caching a much more powerful and high-performance technology than block caching can ever be.
Part II of this blog, resource efficiency, will be posted on Wed Nov 9; part III on management cost, will appear on Fri Nov 11.
Rayan Zachariassen, CTO
November 7th, 2011
Posted by nevex on October 17, 2011
A blog by Nigel Miller, VP Business Development
All of the server-level caching vendors, NEVEX included, will tell you that caching active data to Flash will overcome the performance limitations of hard disk drive storage arrays without the expense and forklift data migration effort required to move an array full of solid state disks.
But there’s more to it than that.
The root cause of your pain point, slow application performance, is caused by the I/O bottleneck. All caching solutions go after the root cause but only one takes that benefit and delivers it with surgical precision to the performance-critical applications that have the most value to your business.
That’s NEVEX CacheWorks.
All of the other caching solutions are block-based and focus on the storage side of the equation. Block caching algorithms specify caching down to the LUN level or storage volume level. Beyond that, they treat all data within the targeted LUNs as having equal value. All data is promoted to or demoted from the cache based on how recently it’s been accessed. Block-based caching algorithms cannot determine the critical value of the data.
NEVEX CacheWorks presents a fundamentally different paradigm – an application-focused view that different data has different value to the organization. Certain applications, reports, and processes translate directly into the dollar fabric of the business.
NEVEX translated its customers’ needs into an application-aware caching solution that allows them to assign priority to data based on its value to the business. Our selective optimized caching algorithms enable administrators to apply almost unlimited granularity in defining data priority to optimally focus the caching on applications that really need the performance boost.
And even further, CacheWorks provides the ability to drill down to select a single database within an SQL Server instance and deliver IOPS acceleration for only the one critical subset that needs the highest possible performance, ignoring everything else within that instance.
Why? Imagine what it would be like if you didn’t have this control…
Picture your business-critical transactional database. It’s currently running well as it’s getting a share of your cache and has primed hot data residing in it. But then someone runs a large and long analysis against your main historics database, and it chews through a quarter TB of data to generate the result. Where’s the share of cache that was accelerating your mission critical database? It’s gone. And with it, any performance gain.
With CacheWorks, your business-critical transactional database wouldn’t be getting a share of your cache; it would be getting all of the cache. And less critical or valuable data would be prevented from causing any collateral cache damage.
CacheWorks is creating a fundamental shift in how caching is perceived and implemented. To date, caching data to Flash has been viewed as a reactive solution to address the storage I/O problem. Our selective optimized caching gives administrators a proactive tool to fine-tune and focus caching on those areas that really need the performance boost. NEVEX is the first and only software vendor to provide this unique and valuable cache functionality.
Nigel Miller
October 17th, 2011
Posted by NEVEX on October 7, 2011
A blog by Steven Lamb, CEO
In previous posts I’ve discussed strategies for a successful technology startup company and what qualities a CEO needs to build and manage a strong executive team. Now that we are ready to unleash our first product, it’s time to share my vision for NEVEX, the value the company brings to the market and the critical customer problems that we solve.
The fact that data storage performance has failed to keep pace with improvements in CPU, memory and networking components isn’t breaking news. But what is not so widely known is that it’s the root cause of the I/O bottleneck plaguing so many companies. And the resulting data delay directly impacts the corporate bottom line, in slower business intelligence, late transactions and general user dissatisfaction.
There are two basic causes of the storage I/O bottleneck. First, hard disk technology. Second, the latency of shared storage with data having to traverse an extended route over SAN or NAS switches, cabling and HBAs to get to the target server. Traditional data storage architectures just cannot deliver data fast enough to meet the needs of the servers running the applications.
Incremental improvements to this problem aren’t enough and the amount of data that companies have to handle is rapidly growing from overwhelming to unthinkable. To give you an idea of the magnitude of the challenge consider that IDC is forecasting that the amount of digital data created will grow from 1.2 million petabytes (or 1.2 zettabytes) this year to 35 zettabytes by 2020 – almost 30 times the current volume.
What’s needed is a conceptual reimagining of data storage that doesn’t simply offer a patch to the conventional outmoded architecture but one that discards current thinking and actually solves the I/O performance bottleneck.
NEVEX bases its ‘data anywhere’ paradigm on the belief that user applications and storage will become increasingly separated in space. We see the solution as being able to present the data content across a cloud of caching devices as a way to provide secure and high-performance data access for business users anywhere.
So how do we accomplish this new storage paradigm? Selective Optimized Caching – more detail on that in a later blog. In the meantime I can tell you that we accelerate applications on physical servers and virtual machines, as well as generic storage I/O. Caching to a server-based Flash/SSD solves the I/O limitations of hard disk arrays and the latency problems of network storage. This architecture allows better utilization of your existing storage infrastructure for capacity and snapshotting while eliminating performance from the primary storage equation.
Our caching solution fully enables the ‘data anywhere’ concept. Your storage infrastructure will soon be able to deliver Flash/SSD-class performance to any application of choice regardless of where the primary data is stored: NAS, SAN, DAS or cloud storage.
In future posts we will detail how our new caching solution operates on both physical and virtual servers, and how it can be performance-tuned for specific applications, data types, locations and users.
Until then,
Steven Lamb
October 7th, 2011
Posted by nevex on August 3, 2011
A blog by Andrew Flint, Product Manager
If there was anyone still debating whether storage performance is really a major concern for business, let me settle it with a simple statistic: each year, companies ranging from medium-sized firms to multinational corporations spend between $15 billion and $25 billion to improve storage performance.
That’s a staggering amount of money, a great deal of which is often wasted in pursuit of solutions that ultimately prove ineffective or inefficient. And yet we still keep looking for answers to our storage woes.
We’ll be outlining how to overcome these challenges in future posts, but for now, let’s analyze the reasons why storage performance issues—in the form of the dreaded I/O bottleneck—gives so many headaches to CIOs and IT managers:
Storage is the slow link in the enterprise. Period. There are three basic elements of performance in a data center; the processing power of the servers, the network speed (switches and routers), and storage, which consists of the disks harnessed by SAN and NAS controllers.
In the past five years CPU performance has increased by 1000%, memory bandwidth has jumped between 500% and 1000% and PCIe bus speed up has grown by 400%. But in comparison, hard drive speed has increased by a meagre 25%. Notice a problem?
A recent blog entry from our friends at Storage-switzerland.com outlines the problem nicely: “Storage performance has not kept pace. Instead it has remained frozen in the same architectural design for at least a decade; a high performance SAN or NAS controller pair that drives an increasing number of disks. While increasing the number of drives can improve performance, there is a limit to the number of drives these controller pairs can support as well as a limit to the amount of inbound traffic they can sustain. This controller (SAN) or head (NAS) is now the primary bottleneck limiting improved storage performance.”
Multi-Tenant Workloads: It’s no secret that workloads are increasing exponentially causing gridlock at the server level. As the aforementioned Storage-switzerland.com article explains: “To compound [the problem of slow storage] the workload is now changing. Workloads are now multi-tenant, with multiple shared servers and networks trying to access storage in this out-dated model. Prior to multi-tenant workloads, a single application coming from a single server could only create a limited number of requests. Multi-tenant workloads, running either through multiple virtual machines on a single physical server or through a single application scaled across many physical servers in a cluster or grid, can now generate hundreds if not thousands of requests for storage I/O.”
More Data. More Data. More Data: The amount of data in the enterprise is growing at an incredible rate. Some analysts estimate that a given business doubles its storage capacity requirements every one to two years! That means more applications leveraging rich content, more data duplication and regulatory storage requirements all continue to consume more and more storage capacity and require additional storage performance.
In my next blog, I’ll outline some of the reasons why CIOs need to pay attention to the I/O bottleneck.
Until then,
Andrew
August 3rd, 2011
