How to Build and Deploy a Well Run Hybrid Cloud #INF3037-SPO w/ @vmtyler & @rick_vmwaretips #VMworld

Notes from the How to Build and Deploy a Well Run Hybrid Cloud #INF3037-SPO session at VMworld: IT has to transform into a service broker (or as I have always thought of it, IT is a service provider).  By working with the business IT should be able to provide the services needed to its users, whether its providing internal/local resources or public – IT can/should be the broker to those resources. emc-hybrid-cloud-reasons Why hybrid?  That is what businesses want; enterprise apps are still being deployed privately – security, control, and licensing are some common reasons.  Building a hybrid cloud allows IT and the business to leverage either resource as needed allowing IT to operate as a service provider (ITaaS) EMC-ehc-pub-priv

 

By enabling ITaaS with EMC Hybrid Cloud for VMware, businesses can expect

Continue reading

VMworld Day 1 – EVO:Rail and lots of vCloud Air

evo-railDay 1 of VMworld, also my very first live VMworld brought some exciting announcements.  These announcements weren’t your typical major vSphere release type announcements, these were much more strategic and for me actually pretty exciting (an hour keynote about 64 node clusters or super-duper-extreme-vMotion would have been a bit of a yawn fest, I expect enhancements like that now).  EVO:Rail and EVO:Rack were the two that were most exciting to me.  EVO:Rail is a solution for VMware partners to deliver hyper-converged infrastructure to their customers.  The boast here is customers will be able to go from zero to working environment in 15 minutes.

Each EVO:Rail hyper converged infrastructure appliance, or HCIA, will consist (at a minimum) of 4 compute nodes with a total of 192GB of RAM (48GB each), 3x 1TB+ 10K SAS drives, 400GB+ enterprise grade SSD and a certified disk controller (which VMware just updated the list of to ensure performance) and 10Gbps networking.  A typical HCIA is expected to support +/- 100 VMs (based of course on your workload).

evo-rail-concept

 

EMC is one of the first EVO:Rail partners (I will be trying to get hands on with one ASAP) along with Dell, SuperMicro, Fujitsu, Inspur, and NetOne.  As you can see above, under the covers this is a HCIA with vSphere and VSAN.  As of today, NSX is NOT included in this configuration (I need to find out if vCNS is or if it is typical virtual switches). EVO:Rail, unlike last  years VSAN announcement, was targeted at partners to enable them to deliver solutions rather than put the burden of installation and configuration on the customer.  I heard a few comments during the keynote along the lines of “there goes my job” but, while simple to deploy, EVO:Rail still needs to be supported and delivered to the customer based on their needs.

I didn’t write about EVO:Rail intentionally yesterday as I wanted to really think about this and determine if I was excited because it was new, or excited because I think it will be a great solution – I am definitley excited because it is a great solution.  Chad Sakacc has an amazing write up here.

Also announced was a rebranded suite of products called vRealize (I realize I’m not good at marketing but man what a horrible name, sorry but it is) that is available as a SaaS based offerings under the recently announced vCloud Air product line (formerly vCHS).  Most notable in my opinion is vCloud Automation Center, now called vRealize Automation for on-premises installs or VMware vRealize Air Automation (at least we don’t have to argue over vCake, vKack, or vSee-A-See anymore).  The SaaS based offering is a somewhat new avenue for VMware.  They had a service called VMware Go but that was cattled a couple of years ago and they still have SocialCast and should be a nice offering for customers that do not have the staff to support onsite vCAC.

All these announcements, however can’t match how amazing it has been to meet so many people in the VMware community, people who have become friends and people I look up to all over social media.  Last night several of us (Jon Harris, Anthony Hook and Gregg Robertson) took in a game at AT&T park.  This, for me is the best part of VMworld.

vmworld-baseball

Introducing EMC RecoverPoint for VMs #VMworld #EMCElect

How do you handle DR? If you are a storage admin you are likely focused on array, LUN or storage pool replication and ensuring all that data is replicated.  However with so much storage, whether local, DAS, SAN, NAS or VSAN you may not need to replicate an entire LUN or array.  Additionally, what if your mission critical VMs span multiple datastores that reside on different LUNs, as they are likely to do for performance and workload reasons?  Keeping track of where each VM is physically stored would be a manual process if you were replicating at a LUN level – imagine storage DRS migrating a VMDK to a different datastore and you weren’t replicating the LUN backing it – you could lose an entire mission critical VM.

Now that virtual machines are “first class citizens” on the array, there is a need to be able to identify and protect at the VM level, not just the the back-end storage.  Today EMC is announcing RecoverPoint for VMs.  The same advanced technology used to protect storage, can now be applied granularly to VMs.

recoverpoint-box

 

vAdmins may be aware of something like vSphere Replication (if its still called that?), and it is similar to that in that both can replicate specific VMs.  RecoverPoint is designed to be easy to install, and manage.  Once installed, you can make changes to the VM such as adding or removing VMDKs from a VM.  Those changes are automatically updated so there is no need to update replication tools.  Additionally, moving a VM to a new host or using storage vMotion/SDRS to migrate VMDKs to a datastore requires no changes – your VM is still protected.

recoverpoint-vms-install

RecoverPoint for VMs will be available in October and can (should) be able to be downloaded for you to try.  It is licensed on a per VM basis with a minimum of 15 licenses required.    You can watch a technical overview here (https://community.emc.com/videos/12056 – note I’ve not watched as the wifis here in my apartment at VMworld is horrible)

Infinio Accelerator v2 announced at #VMworld – Now with iSCSI and FC support

**Disclaimer:  Infinio is a sponsor of virtxpert.com – I was not asked to write this article, I am simply sharing the news about the new features from a great product**

While at VMworld I learned that Infinio has announced a new version of Infinio Accelerator which now supports block storage protocols (iSCSI, FC, and FCoE) as well as NAS (NFS).  Infinio came out of beta last year at VMworld with their RAM based read cache solution for NAS storage.  Now companies who need to accelerate traditional block are also able to do so with the easy to install Infinio Accelerator.

how-infinio-works

From the press release:

In addition to support for the NFS storage protocol, v2.0 will also include full support for Fibre Channel, iSCSI, FCoE, and environments with multiple protocols. No matter which storage protocol customers choose, the Infinio user experience remains the same, including wizard-driven installation, zero operational impact, and instant access to advanced reporting.

In addition to block storage support, Infinio also announced application level reporting and a sizing adviser to help you determine how much memory to allocate to cache.  You can sign up to be notified when the beta of version two will be available at here: http://www.infinio.com/infinio-accelerator-20.  If you are at VMworld this week, swing by booth 623 for a demo.

Infinio-VMworld14-booth-623-map

ICYMI – VMware VSAN Sizing Tool now available

Those nice folks over at VMware have released a web based VSAN sizing tool.  When VSAN was first announced, there was quite a bit of math involved in determine things like usable space, number of hosts needed etc…  Now you can use the VSAN Sizing tool at http://virtualsansizing.vmware.com/; here is a quick walk through.

  • When you first hit the site, click on the Start sizing button (all you anti-web client folks, don’t be scare away by the very web client look and feel)

start-sizing

  • First, enter the information about your virtual machines int he Virtual Machine Characteristics form on the left, for example I used 72 VMs with 1TB (1024GB) VMDKs, with 2 VMDKs per host each using 32GB of RAM.

vsan-sizing-tool

  • Next click on Host Hardware Characteristics (honestly I dislike this part very much, I would have expected the calculator to help me identify the necessary host configuration, I shouldn’t have to tell it what hardware I am using – is that what a calculator is supposed to do?)
  • On the Host Hardware Characteristics page, enter the size of the disks in each hosts, the extra usable capacity (I am assuming GB but you may want to call that out), the number of magnetic (e.g. traditional hard drives) per host and the amount of memory, cores and VM density you would like each host to have.
  • Once that information is entered, the sizing tool will tell you how many hosts you’ll need, the size of flash based cache and other information like total memory in the cluster

vsan-cluster-output

  • In the example output above, I entered 1TB disks, with 100GB of additional usable space with each hosts having 7 disks, 16 cores and 128GB of RAM and a 2-to-1 VM to core over commitment ratio (this is to support a nested production environment – yes I support a nested production environment) so I kept my VM to core low as my 72 VMs are all ESXi.
  • I can see from the output I would need 22 hosts (nicely below the 32 host config maximum) which would provide me with 301TB of total capacity in the cluster.

All in all this is a handy tool, I would like to see a vCPU characteristic added to ensure the cluster size meets the demand of the VMs.  For example without asking how many vCPU my VM has, how can you identify a VM to core ratio which is asked on the Hardware Characteristics page? I would also like to see a host sizing tool, for example I have 72 VMs with 4x vCPU each, 1TB VMDKs and 32GB of memory, what size host do I need (VMware probably wants to avoid customers complaining about performance  if someone uses the tool incorrectly and builds a cluster based on the tools recommendations).

Originally announced at https://blogs.vmware.com/vsphere/2014/08/vmware-virtual-san-sizing-tool.html

What I am hoping for at VMworld 2014

vmworld-featured

**Disclaimer – I am an EMC employee, however I have no insight into future VMware products or features, beyond what is announced by VMware**

I am looking forward to my first VMworld next week, here are a few things I am hoping for more information on:

  • Evo/Evo Rails (formerly known as Marvin):  Marvin is rumored to be a hyper-converged infrastructure appliance – this would be a new path for VMware as it has been strictly a software company.  You can read some rumors on Evo from Christian Mohn at vNinja.net and The Register.
  • Updates on vCloud Air (formerly vCloud Hybrid Service):  vCloud Air is VMware’s IaaS/cloud offering.  In my opinion it operates at a level above commodity IaaS providers such as AWS but up until now you have had to sign up for a contract/commitment.  When vCHS was in beta, there was  pay as you go model – hopefully this is coming back.  The official VMware Press Release can be found here as well as Eric Wright’s take over at his blog discoposse.com.
  • Deep dives in CloudVolumes:  On August 20th, VMware announced the acquired CloudVolumes – a company that seems to simplify application deployment by abastracticing the application from the OS.  Most of the buzz seemed to be from EUC folks (See Kit Colbert’s post here).  CloudVolumes also as a solution for server workloads as well – will this become VMware’s answer to the container craze (e.g. Docker)?
  • Continued advancements for all things vSphere:  While its hard to imagine what “new” features could be added to vSphere at this point (we all know about vVOLs by now), another year of advances in expanding configuration maximums would certainly be welcome.  I for one would love to see an increase in cluster sizes so that I can build larger VSAN clusters and not be limited to 32 nodes.  I’d also like to see continued improvements in their appliance based solutions and a VMware Update Manager appliance.  Though if anyone from VMware is reading this, please for the love of all technology make Geoff Whilmington’s scale out VCSA become a thing!  No SPOF!
  • Community:  What I am probably looking forward to the most is getting to meet so many of the people I’ve had the pleasure of virtually meeting though blogs, twitter and other online social events.  I’ve been able to do it a bit piece meal, getting to meet folks like Cody Bunch, Damian Karlson and Jon Harris but there are so many people I have yet to meet like Trevor Pott, Larry Smith Jr and Alastair Cooke (not to mention so many more).

If you are going to be at VMworld, you’ll likely find me in the community space with the #vBrownBag crew as well as at OpenActs and the VMunderground.  See you on Sunday!

We’ve always been software defined, we just have better software now

Recently I was asked to chat about several topics including Hybrid Cloud, storage and software defined.  It got me thinking about the topic and what I’ve come to realize is, we have always been software defined.  The difference today versus say 10 years ago is we are writing better software and better understand how technology needs to support the business now.

10 years ago, if I purchased a switch for example, from any vendor I would go out and buy something that supported the number of ports I needed as well as through put.  Without software on that switch it would be nothing but metal and circuit boards.  Additionally, the software on the switch would “define” some of the more advanced capabilities.  Depending on the switch, software might unlock features such as routing or firewall like capability.  What we have done today is moved the logic and manageability from the switch up a layer so we can be more responsive to business needs.  It doesn’t mean we weren’t software defined then, it just means we understand better today what needs to be done and can leverage things like NSX, ACI or OpenDaylight to achieve better results.

It’s not just networking, software defined storage solutions are also getting smarter.  We can identify “hot” data and move them to faster drives or take old data and move it off to an archive service all without human intervention.  10 years ago we were probably doing that manually.  We can also use tools like ViPR or Sanbolic to manage heterogeneous storage environments rather than managing each individual piece of hardware.

Now don’t forget, software still needs hardware.  If I have 48x 1Gbps ports that can run at full wire speeds, there is no software in the world that will make those ports act like 10Gbps ports.  The software, if buggy I guess could show them as 10Gbps ports, but I’ll be awfully disapointed in the performance when I start trying to move data through them.  Likewise on the storage side, my software is much smarter, but it can’t turn a 7200RPM SATA disk into an SSD.  I might be able to do something clever like cache it in RAM but its not changing the functionality of the underlying hardware.

What are your thoughts on SDX – marketing gimic or smarter software?

The EMC Hybrid Cloud Solution for VMware

*Disclaimer:  I am an EMC employee.  All views in this post are my own.  I was not asked to write this by my employer, nor was any of this viewed, approved or edited by EMC*

EMC-Hybrid-Cloud-IMG-01Since I started at EMC back in May of this year, I have been fortunate enough to be working on the EMC Hybrid Cloud Solution with VMware.  Cloud pundits might scoff at the idea of having EMC and Cloud in the same sentence, but as I think most would attest you can’t build a cloud without underlying hardware.  In fact you can’t have any “software defined” solution without the supporting infrastructure to run on it whether its Cisco ACI, VMware NSX or EMC ViPR – they call need some form of hardware to interact with.

In case you missed it at EMC World, we built a working hybrid cloud during the show; the goal was to do it in less than 48 hours – it was achieved in a fraction of that.  You can catch the individual demos here (https://www.youtube.com/playlist?list=PLA25JWylP7ffT7kQo2LEzr25eko-RGQs-).  If you will be at VMworld in San Francisco next week, you can come by session 3037 with Tyler Britten and Rick Scherer as they will be talking about how you can leverage best practices to assemble intelligent components to rapidly deploy a hybrid cloud that delivers on-demand services in a multitenant, fully functional, integrated VMware and EMC solution stack.

Continue reading

Ansible (@ansible) releases 1.7 with beta support for Windows

image-ansibleAnsible has announced version 1.7 which introduces beta support for Windows systems.  Ansible, a configuration management tool, has long been a very powerful tool to manage configuration on Linux systems and now that is being extended to Windows (http://docs.ansible.com/intro_windows.html) through PowerShell Desired State Configuration.

I was introduced to Ansible at my previous company where a co-worker was able to orchestrate and work magic with this tool.  I was able to learn quite a bit just from looking over her shoulder and absorbing everything I could from her.

While 1.7 supports Windows as a beta feature, Ansible also announced that Windows will be fully supported in 1.8

Ansible, and other configuration management tools such as Puppet and Chef have introduced a new way to think about how you manage your systems, at least for me as former Windows administration.  Along the lines of “pets” and “cattle” – you don’t care what the current state of your system is, you only care about what the final state is.  This tool falls in line with my personal beliefs that standardization is a key tenant of IT and ITaaS which helps organizations be secure.  After all if your systems are not standard, you can’t validate that they are secure.  Ansible helps you achieve that standardization by applying configuration or settings.

Ansible comes in both a paid and a free version for up to 10 hosts, so there is no reason not to get started with Ansible!

Quantum (@QuantumCorp) Acquires Symform

quantumWho and who you say?  Remember back in the 90′s….Quantum, those folks that made a hard drive you would consider after Seagate and Western Digital?  Yea those guys!  Well guess what they are still around and have some interesting backup options in the SMB space.  In fact, they offer their backup software – vmPRO free up to 1TB and a virtual version of their backup and dedup appliance free up to 15GB.

What Quantum has lacked is a cloud service offering – enter the Symform acquisition.  Symform makes backup and file access software that is available across multiple devices.  Based on the press release, it sounds like Quantum is looking to integrate this into their existing offerings to have a cloud enabled solution.  Imagine running their vmPRO backup software to protect your VMs to a local NAS device or their virtual DXi V1000 AND then also being able to create offsite backups directly from the software or appliance.

From the press release:

Symform is based on software that allows data to be stored transparently in more than one location. Today this is just a horizontal cloud, but this capability allows Quantum to potentially expand the service to include directing one copy of a customer’s backup to stay onsite and a second copy to the cloud, in a hybrid onsite/offsite, but fully managed model.

If you are evaluating VM backup software, its certainly worth putting on the list.  I do wish they made registering for the free software a bit easier, I had previously registered but lost the key and when I go back there isn’t a clear way to retrieve it, nor can I register again.  They’d also benefit from being a bit more social, I had trouble finding their Twitter account!