IOPS Per Dollar

Usually we do GB/$ when it comes to buy storage, we want storage space, huge storage space for less money… But sometime we need fast storage, read lots of IOPS, and then comes in the idea of IOPS/$. Usually we quote for 15K disk, in RAID1+0 for maximum performances. Such configurations cost a lot of money! Did you know that even thought SSD (Solid State Disk) are much expensive per gigabyte  than a 15K, it is much cheaper when talking of IOPS/$.

Chad Sakac just posted a very interesting article on SSDs. He was a bit skeptical about this new technology but when looking at the figures, he changed his mind.

A 1.5TB 7200 SATA drive can be bought for $115.   But it will do 80 4K random write IOps. So:

• Expressed as GB/$ = the SATA drive is a 38x better deal
• Expressed as IOps/$ = the SSD drive is a 50x better deal (for write IO workloads, for reads, it’s 93x better).   The SSD delivers 34 IOps per dollar.   The SATA drive deliver 0.69 IOps per dollar.

To match the random read IO performance of that $115 Solid State disk, you would need 50 of the 1.5TB SATA disks.

Another article, VMware vSphere^TM 4 Clocks 364,000 IOPS in Record Storage Throughput, demonstrates the performances of SSD disks 6 months ago already.

there Is More To Desktop Virtualization Than VDI

Sure there is more!  Can you tell me what are the other versions?

I was reading Daniel Feller’s blog post and wanted to better understand the other options he was talking about. In Daniel’s post there is a link to a CIO.COM article, bingo, that’s what I was looking for… This is the far most interesting article regarding Desktop Virtualization I have ever read, Kevin Fogarty, the author detailed his article with pro’s and con’s, also giving vendor offerings.  Let me share it with you! 

With all of the market competition around desktop virtualization, some users are confused about the pros and cons of the various options. Here’s a snapshot of the major desktop virtualization approaches and of the types of situations for which they might be appropriate. 

1. Remote Hosted Desktops:

What most people think of when they think “terminal services.” A server runs one image of an operating system or application and many clients log in to it using connection broker software that is the only part of the software hosted on the client machine. Client machines operate only to show an image on the monitor of the application that user is sharing, and to transmit keyboard and mouse input back and forth.

Advantages: Low cost, high degree of control over data and applications.
Disadvantages: Performance depends on the quality of the network connection; display protocols often can’t handle complex graphics; some applications designed for desktops can’t run in shared mode on the server; inflexible for end users, who can’t store data locally, use most peripherals, or move data back and forth using thumb drives. Does not work when disconnected.

Example vendor offerings:
Software: Citrix XenDesktop; Wyse ThinOS; Microsoft Remote Desktop Services; Microsoft Enterprise Desktop Virtualization (MED-V); VMware View Manager.
Hardware: Pano Logic Device, Remote; nComputing thin clients; Wyse thin clients; Sun Ray Ultra-Thin client; Symbiont Network Terminal; Rangee Thin Client 

2. Remote Virtual Applications:

What you get in every Web application you’ve ever used. Differs from shared desktops in that the only thing required is a browser and standard Web protocols (HTTP, HTTPS, SSL, etc.) to create secure connections and transmit graphics and data. Depending on design of the applications (think Flash downloads) the end-user’s machine may process some of the application’s logic or graphics, or may only light up the monitor and send clicks to the server.

Advantages: Doesn’t require that IT control the hardware or software environment of the end user.
Disadvantages: Doesn’t allow IT to control the hardware or software environment of the end user, which could affect performance. Does not work when disconnected.

Example vendor offerings: Citrix XenApps; Microsoft Remote Desktop Services; VMware View; VMware ThinApps. 

3. Remote Hosted Dedicated Virtual Desktops:

The next step up in power for end users and step down in cost and resource conservation for IT from Web apps or terminal services. Rather than having many users share one instance of the same application or operating system, the server hosts an entire operating system and set of applications within a virtual machine that is accessible only to that user. The VM could run on a server, sharing resources with other dedicated VMs, or could run by itself on a blade PC. Can either be hosted remotely or streamed. In the streamed scenario, both applications and operating systems can be streamed to the client—downloading parts of the software as the user requires them, and executing on the client machine, using its processing power but not local storage. 

Hosted remotely:
Advantages: Can run applications that balk at running in shared mode; isolates activity of each user to prevent resource constraints.
Disadvantages: Uses far more bandwidth than shared desktops, and far more hardware on the server. Performance still depends on the quality of the network connection and ability of the display protocol to handle graphics. Does not work when disconnected.
Example vendor offerings: Citrix XenDesktop; Wyse ThinOS; VMware View; Microsoft Remote Desktop Services; Microsoft Enterprise Desktop Virtualization (MED-V) 

Streamed:
Advantages: Often gives the end user better performance because demanding graphic or other operations execute locally.
Disadvantages: Requires more powerful client hardware, reducing the cost benefit of virtual desktops. Does not work when disconnected.
Example vendor offerings: Citrix XenDesktop, XenApp, XenProvisioning; Wyse TCX; VMware View Manager, ThinApps, Composer; Microsoft VDI suite. 

4. Local Virtual Applications:

Think “Java.” Applications download from the server to the client machine and run there, using local memory and processing power. But they run within a “sandbox” that enforces a set of rules on what the local machine can do and to what it can connect.

Advantages: More computing resources and sometimes better performance than remotely hosted applications; less bandwidth consumption; can be used offline.
Disadvantages: Less control by IT over the hardware and security of the data.

Example vendor offerings:  pre=”">Citrix XenApp, Wyse TCX, VMware ThinApp, Microsoft Application Virtualization. 

5. Local Virtual OS:

Present in two major versions. Option one: A client-side hypervisor can create a virtual machine within a laptop or desktop computer, which can function as a completely standalone unit that keeps itself separate from hardware and software on the client machine outside of the VM. Option two: A hypervisor runs on the machine’s BIOS, allowing the user to run multiple operating systems with no “host” OS at all. 

Advantages: Multiple OSes on a single system; no concerns about OS compatibility, can run on non-traditional VM clients such as smartphones or PDAs.
Disadvantages: Potential conflict for resources, relative immaturity of client-side hypervisors leaves security unproven.

Example vendor offerings: Citrix Dazzle and Receiver, Wyse PocketCloud, TCX, VMware View Client Virtualization with Offline Desktop (Experimental); Microsoft VDI suite 

 

Sources: Citrix.com, Cio.com

Formula To See How Much You Save On Physical Memory With VMware PSHARE

Part of the overall memory overcommit feature, VMware has this cool feature called transparent memory page sharing (TPS).

When ESX Server detects an extended period of idleness in the system, the VMkernel begins to compare physical memory pages using a hashing algorithm. After encountering two memory pages that appear to have the same contents, a binary compare is executed to ensure similar content. ESX Server then frees up one of the memory pages by updating the memory mappings for both virtual machines to point to the same physical memory address. Should a virtual machine attempt to write to or modify a shared memory page, ESX Server first copies the shared memory page, so that a distinct instance is created for each virtual machine. The virtual machine requesting the writte operation to the memory page is then able to its contents without affecting other virtual machines sharing this same page.

With memory sharing, a workload consisting of multiple virtual machines often consumes less memory than it would when running on physical machines. As a result, the system can efficiently support higher levels of overcommitment. The amount of memory saved by memory sharing depends on workload characteristics. A workload of many nearly identical virtual machines may free up more than 30 percent of memory, while a more diverse workload may result in savings of less than 5 percent of memory.

Now that we have seen the definition, let’s see how to calculate this number. First you need to install vSphere PowerCLI available here, then you need the Get-Stat2 cmdlet available here. Once you have all these in place and installed, fire up vSphere PowerCLI and browse to the directory where you have the Get-stat2.ps1.

Now here is the formula:  

[math]::round(((Get-View  class=”hiddenSpellError” pre=”Get-View “>-ViewType  ”>HostSystem |?{$_.Runtime.ConnectionState -eq “connected”} | %{((((./Get-Stat2.ps1 -entity $_ -stat “mem.shared.average” -Interval RT) | measure-object -Property value -Average).Average) – ((((./Get-Stat2.ps1 -entity $_ -stat “mem.sharedcommon.average”-Interval RT) | measure-object -Property value -Average)).Average))}) | Measure-Object -Sum).Sum /1Mb,2)

Personally my output is 272GB. That is the sum of memory page shared among all my VMs across a bunch of hosts. That’s huge and knowing the price of memory for servers nowadays, I just can tell this features is a must have and so far only VMware has it since … 2001 WOW!

Let’s talk money a bit. I’m running HP blades, BL460’s mainly, 272GB of HP memory is about 80,000Eur knowing that MSRP price list for a 2*4GB PC2-5300 kit is about 2,350Eur , no VAT, no rebate.  With that saving, I paid my VMware licenses!

You can follow real-time the gain by using the performance tab of your vSphere Client. I will post a meaningful picture later. The former one is wrong. Thx to Hypervizor for letting me know

Sources: Hypervizor.fr, Windowsitpro.com, VMware.com

Thin Provisioning A P2V, How Do You Do That?

Yes indeed, whilst you convert, you can resize the target disks but hey, as Rick Vanover experimented, it is twice as slow as doing a regular block level copy. Why? because when resizing the target disk, vConverter switches to file level copy and it’s really slow…

Now what’s the trick? The issue with the faster process is that there is potentially large amounts of wasted space on the SAN. This can be solved by vSphere’s Enhanced Storage VMotion. The enhancements in vSphere allow you to perform a storage migration task from a fully allocated virtual disk to a thin-provisioned disk. This means that once the workload is converted, you can perform this task to reclaim that wasted space. This is a timesaver, as the Storage VMotion task is done online with the virtual machine. The only caveat is that there needs to be enough space on the storage platform to support all of the pieces in motion, which may add up if there is a lot of free space involved.

Thanks Rick for the trick!

Source: Virtualizationreview.com

Trust Me!

EMC PowerPath/VE, Get It For FREE!

Eric Sloof just published a very interesting article about PowerPath/VE. Following up with one of my previous post about the same subject, I would like to add some info and summarize what does PowerPath/VE compared to ESX’s native multipathing.

First we must understand 2 definitions:

-Multipathing: maintains more than one physical path between the host and the storage device. If any element in the SAN fails, or at the host, such HBA, cable, switch, the host can switch to another physical path that DOES NOT use the failed component. This process is called path failover.

-Load balancing: is the process of distributing loads across multiple physical paths.

ESX4 natively supports multipathing (NMP) only, no load balancing functions here. Actually ESX4 and former versions, see the storage device in an Active/Passive way. Only one path at a time is used to send data from the host to a storage device, so basically only one path out of your 2, 4 or more paths is used. Failover is triggered by one of the 3 available policies; fixed, most recently used path and round robin. N.B. round robin is meant for production in vSphere only (ESX4).

To add load balancing support, you must install a Multpath Plug-In (MPP) that uses the vStorage APIs for Multipathing (only available in vSphere). EMC PowerPath/VE is the first to offer that plug-in.

PowerPath/VE works with VMware ESX as a Multipath Plug-In (MPP) that provides path management to ESX hosts. It is installed as a kernel module in the vSphere host (ESX host). It will plug in to the vSphere I/O stack framework to bring the advanced multipathing capabilities of PP/VE including dynamic load balancing and automatic failover to the vSphere hosts.

In few words, it enhances default ESX’s multipathing function and add dynamic load balancing to vSphere.

Now the cherry at the top of the cake, you can trade in your physical PowerPath licenses for PowerPath/VE licenses in most cases, so it can be very low cost (in some cases free).

Introducing Windows Server 2008 R2 eBook Available as Free Download

Introducing Windows Server 2008 R2 is a deep-dive work, that’ll get you up to speed on how R2’s new features and capabilities work, including Hyper-V and RDS virtualization, management, IIS and the new Web application platform and, of course, all the synergistic goodness between Windows Server and Windows 7. Free for a click, so don’t miss out.

Source: Technet.com

Dell attempts to copyright ‘cloud computing’

The company is applying to trademark the term “cloud computing,” according to a document on the US Patent and Trademark Office’s website.

The application has reached the Notice of Allowance phase, where a company receives “a written notification from the USPTO that a specific mark has survived the opposition period … and has consequently been allowed for registration,” according to the USPTO website. The opposition period gives other parties a chance to object to an application.

That reminds me IBM and its Personal Computer aka PC… 

More at Techworld.com

Cloud Computing! It’s worse than stupidity: it’s a marketing hype campaign

Richard Stallman, founder of GNU foundation says it loud and clear, the concept of using web-based programs like Google’s Gmail is “worse than stupidity“.

He says cloud computing was simply a trap aimed at forcing more people to buy into locked, proprietary systems that would cost them more and more over time.

“Somebody is saying this is inevitable – and whenever you hear somebody saying that, it’s very likely to be a set of businesses campaigning to make it true.“

“The computer industry is the only industry that is more fashion-driven than women’s fashion…“

“One reason you should not use web applications to do your computing is that you lose control,” he said. “It’s just as bad as using a proprietary program. Do your own computing on your own computer with your copy of a freedom-respecting program. If you use a proprietary program or somebody else’s web server, you’re defenceless. You’re putty in the hands of whoever developed that software.“

Source: Guardian.co.uk

TwInbox, a Twitter Add-in for Microsoft Outlook

If you are a Microsoft Outlook user, you probably have it open all the time. Now you can have a fully-featured, powerful Twitter client at your fingertips without having to open any other applications. TwInbox seamlessly integrates Twitter into Outlook. It is a perfect Twitter client for any Outlook user, from a Twitter newbie to a business professional.

Go and get it at techit.com