Virtualize Business Critical Applications (blog) review

30 Jan

So this I guess would be a meta blog.  The source is here.  I am glad to say that after the Conceptual, Logical, Physical reading this one made way more sense.  This blog goes into how to address the fears of virtualization, mostly by application manager.  VMware has won over C level people by showing a cost savings from the old days of a server, or cluster of servers per application.   It won over operations by simplifying DR and reducing cabling, and electric consumption.  It won over Network and Systems Admins with the simplicity to mange, and greater agility.  The last hurdle is application teams.  Their concerns a broken down into performance, scaleability, availability, supportability, agility and disaster recovery.

Performance wise the blog goes into studies which show that on average at most there is a 5-6% hit on performance.  IT breaks this down by SQL which performed almost as well as native, SAP which showed a 5-6% drop, exchange cluster which showed about a 5% hit, and oracle rac.  So overall this should be able to be used to offset any performance concerns.  The trade up of being able to refresh, your workload onto new hardware within days instead of weeks or moths should be worth the 5% overhead.  Also I would be concerned with a system which runs constantly hot at over 90% utilization for over 50% of its life-cycle.

Scaleabilty has been addressed since esxi 5.5 with the introduction of vNUMA.  NUMA is non-uninformed memory access.  So what this does is provide a virualization layer of the physical NUMA architecture which is on the underlying hardware.  This also allows for operating systems and applications which are NUMA aware to be able to take advantage of this when scaling up.  This is combined with the hot add feature of vRAM and vCPU.  So hot add aware OSes and applications know they are running in a system with 2 vCPU and 4GB of vRAM.  They also would be able to tell when VMware hot added and other 2VCPU and 4 GB of vRAM providing a total of 4 vCPU and 8 GB of vRAM.  This is a major advantage for workload and application which are hot add and NUMA aware.  Combining that with the new configuration max per VM of 64 vCPU, 1 TB of vRAM, up to 40GB/sec networking and up to 1,000,000 iops on storage mean that there is almost no workload which VMware could not scale up to.  VMware already had scale out down with making it much simpler and faster to deploy another VM to add capacity.  This is all with proper underlying hardware of course, and I guess that is the trick we are trying to learn.

Availability is the next question.  VMware started off with HA in ESX 3.0 I believe.  This was a simple feature which monitored all of the VMs, when a host went down vCenter was aware of which VMs were on that host and then restarted them on another host.  This was a first step, but as up-time demands increased simply having the server back after a quick reboot was not enough for all workloads.  Another feature of Fault Tolerance was introduced in ESX 4.0 this allows for a basic shadow VM to be running on a different server in lockstep.  If this initial server when down this shadow server was promoted.    There are many limitation to this but with ESXi 6 you can now have up to 4 vCPU depending upon licensing, in version 4 and 5 you could only have 1 vCPU.  This is an improvement but for your true HA ideally you will want to cluster on top of a VMWare cluster.  You do need to watch this with running VMware HA on a software clustered work load.  Typically anti-affinity rules are used to make sure that each VM in a virtualize cluster are running on different physical hosts.  This can cause problems with vMotioining things around for maintenance.  I like to keep all of the applications which are in a virtual cluster, on one VMware cluster if possible.  This leave that type of work load in one place.  Optionally on this cluster you may not even need to enabled HA since all of the workloads are theoretically able to lose nodes without being affected.

Manageability has been greatly improved with augmenting vSphere with vCOPS.  vCOPS allows for predicative analysis of use as well as near real time monitoring.  It makes is easier to coralate some of the data and to see where an issues root cause it.  Combdingin this with Log Insight also helps out with having a possible single repository for all log files to find root causes.  It does add another layer of administration, however.  Alerts cleared in vCenter do not automatically clear in vCOPS so while it is great for larger environments there is room for improvement on two way communication between vCetner and vCOPS.

Supportablity used to be a bigger issue, with certain applications saying they did not support VMware virtualization.  That has pretty much become an issue of the past thought.  It is still recommended when designing to engage to make sure that there are no particular configurations which may not be supported.

Agility has been a strong point of ESX from the begging.  With the ability to create customized templates of applications Test, Dev, PrePRod, and Prod workloads can all be spun up faster.  Test, Dev and Pre-Prod can be destroyed and redeployed and restored in much less time leading to much faster development cycles.

Disaster Recovery is the final point to address and this was met with a separate product call Site Recovery Manager.  SRM uses work flows to plan out what will happen in case of a disaster.  It utilizes either vSphere replication or array based replication as the means by which the data is replciated over to the recovery site.  It also runs automation in order to test, or fail over a workload.  It takes some planing out, but it is a very versatile tool to implement.



Conceptual, Logical, Physical: It is Simple

8 Jan

This has to be one of the hardest reads, I have tried to get through and it is only 7 pages.  The link within the guide takes you here.  Which is another post from John Hall from October 2011.  The actual document which is referenced is originally here.   As is noted within the comments of this post the actual matrix which is being referenced is not in the doc.  There is one link within the VMware post here.  I have also used this wiki link here.

The wiki may be a better read, but on to this.  Page 1 is an example with the end pointing out that as Architects, who are in row3 a common issue is that they confuse the entities which are in their row for things which are in row 2 or the Business Management perspective.  And that entities are simply a logical representation of business things.    Or that a common mistake is to have the Architect name the Business Requirements “things”.  So row 2 is the models of how the business management perspective.  Row 3 is the Architects logical implementation of row 2

Page 2 goes into why row 3 is important not to skip or mix up with row 2.  They point out that you don’t build things via trial and error, that is expensive.  So you sketch out on paper.  It also points out that if you are trying to work with an existing system you need to start at row 6 and reverse engineer everything back.

Page 3 goes into the second reason why row 3 is important, and that is as a business grows the gap between what the Business Management Perspective of Row 2 is get diluted as they can no longer have physical access to everything.  He then defines surrogates which are created as blue and white collar.  At the end of page 3 it is revealed that this is row 4 or the engineer perspective.  And based on the system can be either blue or white collar or maybe it can only be blue collar, I really am not sure.

Page 4 maybe clears things up a bit in the fact that they kind of say that Conceptual, Logical, Physical, are set to Row 2 Row 3 and Row 4.  So Maybe if I am understanding any of this Row 2 is how a thing is thought to be, row 3 is a paper model white collar of how that this is, and row 4 is the physical model or implementation.

In Page 5 there is a concept that I clearly get, which is that the Quality of the cell you are in needs to be defined by the cell above it.  Most basically does the cell you are producing meet the requirements of the cell above it correctly, and does it get successfully implemented in the cell bellow.

In Page 6 things clear up a bit in the fact that while they keep talking like this is a 2d model it seems more clear that is a 3d model.  So Row 2 is The Owners View, The Conceptual Enterprise Model, The Business Process Model and the Business Managements Perspective all at once.  It also goes into some “Absolutes” which is defining Row 2 as Models of the business Row 3 as Models of the Systems, and Row 4 as Technologically constrained Models.  I guess I have spent most of my life on Row 4 and never even knew it.

To give anyone reading this an idea.  I had to read the pdf 5 times, I had to go onto that wiki page and read that 2-3 times and between the two I may have made a little sense of this.  Will any of this be on the test?

Functional versus Non-functional Requirements

6 Jan

This is a link here to a VMware community post by John Hall from 2011.  This 4 page PDF defines the diffidence between functional and non-functional requirements.

Functional requirements – a specific function that the system or component must be able to perform.  Examples being Business Rules, Transaction corrections, adjustments and cancellations, Administrative function.

Non-Functional Requirements – specify criteria that judge the operation of a system, rather than specific behaviors.  Performance response time, Capacity, Recoverability ect.  The are quality characteristics or quality attributes.  Noting it is wise to have many different stake holders present for this portion of the definition.

I find it interesting that the document gives many examples of functional and non-functional requirements, but does not dig in to their meanings.

The doc then goes over the Traceability matrix.  It goes over Functional testing, which is more of a yes no test.  Either the system can or cannot do functional things.  These are not quantifiable.  IE it either does or it does not have the ability to vMotion over from one server to another.

Then it goes into Non-Functional testing, which is more of the testing that I am used to.  Performance testing, Usability ect.  IE how quickly can the system vMotion over a VM with 64GB of RAM from one host to another.

The final thing which is defined is the V-Model which allows for testing and “coding” which I would feel would be more design in this case to occur in parallel.

Overall most of this doc contains things that I never really though of.  It was a quick and easy read.

Review of The Benefits of Virtualization for Small and Medium Businesses

5 Jan

This doc is a bit dated with it being published in 2009.  The executive summary on page 3 could be summed up as… SMB like virtualization but their budgets tend to keep them from expanding upon the virtualization.  This is the case even though the benefits of virtualization to reduce risk of data loss and it outages is known.

The benefits of virtualization on pg 4 goes into the results of their survey and covers the basics of why virtualize.  Improve server utilization, reduce server footprint, improve security, improve availability and up-time, improve application management and development cycles, and improve backup protection and recoverablity.

One interesting stand out was the fact that IT decision makers and business decision makers seem to not agree on how responsive IT is to the businesses needs.  Not that this is unexpected, just funny to see in a white paper. Page 5 delves more into this pointing out that companies which virtualize have a better perception of responsiveness.  Well if you can deploy a server in a day(s) instead of weeks, of course you will be able to be more responsive.  Maybe this was less understood in 2009 and a harder fight.

Page 6 goes into Data Loss numbers and outlines how there tends to be a disconnect between the business and the IT dept on how good their DR is setup.  An interesting point was that they highlight the average cost which companies assign per hour of down time as 23,188.00.  I am very interested in how they came up with this number.  Of course VMware drastically improves recovery time, by making it much simpler to recovery your servers on different hardware.  That is nothing to mention, being able to use SRM to have a hot/warm/cold recovery site, or using VEEAM to have full backups of your VMs which are very easy to recover vs traditional server recovery.

Page 7 goes into why SMBs say they have not virtualized yet with lack of budget being the number 1 factor, and the number 2 factor is not having the proper resources.  This is of course why there is so much interest in these certifications 🙂

Page 8 goes over IT Budgets and Goals.  I am guessing that this is just an outline of things that should be asked to Stake holders to make sure that you are have a good picture of what a business wants and how VMware can help them get there.

Page 9 goes over a recap of the benefits of virtualization.  There is also a small getting started with virtualization which highlights the free version of ESXi and also shows the VMware ROI calculator.

Page 10 breaks down the methodology which I find fairly interesting.  All in all I would say not the most interesting or informative thing to review, but this was one quick way to get back to studying every day.


VMware Certified Advanced Professional 6 – Data C enter Virtualization

5 Jan

It has been quite some time since I have posted a new blog.  I am writing this up mostly for myself, to keep track of the studying which I am doing for the VCAP6-DCD or VCIX6-DCD as VMware is calling it now.  Joshua Andrews was kind enough to post the Beta blue print on linked in.  I will be using that as the start of my study guide.

I am going to break down each of the Objectives which are listed.  I will be then doing a separate post on each fleshing them out as I learn more about what will be focused on within the Exam.  To start off I will be reading all 54 pdfs which are mentioned and I will highlight stuff I find interesting as well as what I find within other research to be what the test will be interested in.

My goal is to try and take the exam before the end of April.  A possible issue with this is the fact that VMware still has not even finalized what is going to be on the exam.  Still there is no place to start like the present.

Objective 1.1
Gather and analyze business requirements




Functional vs. Non-Functional Requirements

Conceptual, Logical, Physical: It is Simple.

Virtualize Business Critical Applications (blog)



Lab Build Section 5 Creating AD Controller on the Base Server Part 1 Create VM

12 Feb

Goeie dag… and welcome to my blog on creating an VMware ESXi Lab for certification training.

In this section we have been going over Creating AD Controller on the Base Server.  In this Part we will go over Creating a VM.  Due to the functionality of documenting the process of creating, this post will be re-referenced many times latter in the process of documenting the development of the lab.

To start off you will need to log onto your ESXi host directly.  If you are unfamiliar you can follow this link.

Once you are connected to the ESXi host via the vSphere Client we will need to bring up the New Virtual Machine Wizard.  This can be done several ways.  The quickest is to hit CTRL + N, other options are to go to the file menu select new and then select Virtual Machine, you can also right click on the host and select New Virtual Machine…

New Virtual MachineThis will open up the Create New Virtual Machine Wizard.  From the Configuration screen there are two options, typical and custom.  If you select Custom you have extra stages set in the wizard for setting Virtual Machine Version, # of CPUs, Amount of Memory, and type of SCSI controller.  For this deployment we will use Custom to go over these options.  Select the Custom radio button and click Next.

New VM Wizard Configuration  At the Name and Location screen type in a descriptive name for your VM.  I would highly recommend coming up with a naming convention in order to keep your VMs logically named.  This was covered in previous blog posts.  In this example EP-W2k8-DC is being used.  Click on Next.

New VM Wizard NameAt the Storage Screen you will see the options for the locations to store you VMs.  You can see that the datastore also follows some naming conventions with the Site-Device-Storage Type Configuration.  To start off with we will place this VM in our Direct Attached Storage. Click Next.

New VM Wizard StorageAt the Virtual Machine Version there is the option of selecting what version to create the VM with.  It is best practice to select the highest version that your environment can handle.  Since this is a lab there is no need to worry about mixing version with old versions of ESX or ESXi so in the example Virtual Machine Version 8 will be selected.  Click Next.

New VM Wizard Virtual Machine VersionAt the Guest Operating System screen, there is radio buttons for Windows which is default, Linux and Other.  Based on this the Version: drop down will be populated with choices.  In this example the Windows Radio button is selected and the Version is Microsoft Windows Server 2008 R2.  Click Next.

New VM Wizard Guest Operating SystemAt the CPUs screen there is the option of selecting how many virtual sockets will be used, as well as how many cores per virtual socket will be used.  In this example the default values of 1 will be kept.  Click Next.

New VM Wizard CPUs

At the Memory screen there is the option to select how much virtual Memory will be assigned to the VM.  In this example the default of 4GB is kept.  Click Next.

New VM Wizard MemoryAt the Network screen there is options for how many virtual NICs to assign to the VM.  After the number is selected from the drop down each vNIC needs to have it’s network label selected from the drop down as well as the adapter type and a selection if the vNIC will be connected at Power On.  The Network label denotes the virtual network that this vNIC will reside on.  This will be covered in greater detail in latter sections of this blog.  When choosing the adapter type a good refernce is the following KB posted by VMware

This example uses one vNIC attached to the default VM Network with an adapter type of  VMXNET3 that will be connected at Power On.  Click Next.

New VM Wizard NetworkAt the SCSI Controller screen there is 4 options for SCSI controllers.  The default is selected based on the Guest Operating System that was chosen.  In this example LSI Logic SAS will be used.  Click Next.

New VM Wizard SCSI ControllerAt the Select a Disk screen there are 4 radio button options for types of disks to use.  Depending upon the radio button selected, the wizard will guide through the next steps.  In this example the default of create a new virtual disk will be used.  Click Next.

New VM Wizard Select a DiskAt the Create a Disk screen, there are several sections.  The first section is the Capacity.  Here the disk size is set.  The second section is Provisioning.  There are 3 options, Thick lazy zeroed, Think eager zeroed, and Thin.  There is an in depth article on provisioning here.  The third section is location.  There is the option of Storing the disk with the Virtual Machine which is default, or the option to specify a location.  The reasons for choosing different locations will be covered latter.  In this example the Disk Size has been changed from the default 40 to 60 due to the amount of Windows updates that have been seen on Server 2008 R2 filling up the C drive.  All other options are kept as default.  Click Next.

New VM Wizard Create a DiskAt the Create a Disk Advanced options there is two sections the Virtual Device Node and Mode.  The Virtual Device Node allows you to define where this drive will be on the SCSI or IDE logical chain.  The Mode has advanced options that will be reviewed at a latter time.  In this example the defaults will be kept.  Click Next.

New VM Wizard Create a Disk Advanced OptionsAt the Ready to Complete screen there is a listing of all of the settings.  This is a good place to review all choices and validate that there are no mistakes.  Once completed, click Finished.

New VM Wizard Ready to CompleteOnce the finished button is hit a Create virtual machine task will pop up in the recent tasks.

Recent Tasks  Once this reaches 100% the virtual machine is created.  The next part will cover installing Windows Server 2008 R2.

Lab Build Section 4 Installing ESXi on a Base Server Part 4 Connecting to the Host via the vSphere Client

12 Jan

Hallo… and welcome to my blog on creating an VMware ESXi Lab for certification training.

In this section we have been going over Installing ESXi on a base Server.  In this Part we will go over Connecting to the Host via the vSphere Client. I have decided to break out this section into such a small sub section, in order to be able to easily reference it in latter post.  There really is nothing to it:-)

To start off you will need to log onto your ESXi host directly.  This is due to the lack of vCenter servers yet.   You will need to know the ip address of your ESXi host, as well as the root password.  All of this information was set in previous posts.  To do this you will need to click on your vSphere Client Link on your computer.  vSphere Client LinkThis will open up the vSphere Client.  Fill in the correct IP address of you ESXi host.  Note we currently cannot use the name because we do not have DNS setup yet.  Also fill in root for the UID and fill in the password for the root user.  Click on Login.

vSphere Client LogInThis will bring you to the Getting Started Tab of the vSphere Client.  This is the jump off point for much administration.

vSphere Client Getting Started