Flipping the Cloud Paradigm – part II

A little over a year ago I wrote “Are you a cloud paranoia”. The baseline of the post was that people don’t trust Public Cloud – mostly rightfully so – to own all their business information. We briefly touched on it in last week’s episode of In Tech We Trust as well due to Microsoft Azure (blob storage) going titsup again for more than 11hrs.

But it’s the last part of that blog I still think of a lot. I proposed the idea of flipping the Cloud Paradigm and running your total production in (public) clouds but using your local infrastructure as your DR-location. Here were some of the reasons: 

  • I still own my data when sh*t hits the fence
  • My buying cycle shifts almost completely from
    CAPEX to OPEX
  • My own infrastructure including power, cooling
    & hardware maintenance is minimal
  • If my own infrastructure is minimal maybe even
    my staffing could be minimal (this cost will shift to well trained service
    providers)
  • Public/shared cloud providers have a core
    business is making their infrastructure as efficient as possible (SuperNap/RackSpace/…).
    If your provider adds new services that didn’t exist last year you’ll be the
    first to benefit from that.
  • you probably will have a better relationship
    with some consultants when you need help/advice Continue reading
Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

My checkbox is bigger than your checkbox!

This morning Josh Odgers from Nutanix published a blogpost where he rightfully pointed out that having support for a specific feature(set) doesn’t necessarily mean the same for all vendors (link). In this case we are talking about NAS-VAAI. The example given is that there are a lot of vendors that have support for NAS-VAAI but Nutanix is the only one that supports, and is certified for all primitives. Point well taken Josh and luckily there is always the HCL to check that out as you showed.

note: the same applies to VAAI for block primitives!

Of course when one rep from a vendors says A, others will say B and then we get good technology discussion on twitter. Sometimes with trolls and dragons but in the end we all learn if we are open to it. And that’s when this spark came along. 

All checkboxes are created equal but some are more equal than others.

I have seen one too many sales-reps (and customers as well) doing checkbox sales and acquisition. Do you have this feature? YES, so we must be better than all the others. WRONG! And I’ll show you with XCOPY as an example.

Screenshot 2014 11 28 09 51 02

This is a plain VM copy within the same host, to the same datastore. What happens is that the ESXI host reads every IO and then copies it back to another location. This is how we all did up till we got VAAI.

Continue reading

Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

The horror of Agent-Based system restores

A day in the life of a Data Protection Auditor

STEP 1: List VM hardware details!
  • vCenter (CPU/MEM/vDISK/vNIC)
  • including vNIC & SCSI Controller type!!!
STEP 2: List in-Guest Backup Agent details (case-sensitive)
  • http://backup.local
  • Remark: Operating Systems description are not (always) up to date, which could lead to wrong choice of template.
  • Remark: Everyone using the same (personalised) admin accounts is not a best-practise for logging purposes > use admin groups!
STEP 3: list IP Configuration details
  • http://cmdb.local
  • Issue: NO DNS information available
  • Remark: everyone using the same (personalised) admin accounts is not a best-practise for logging purposes > use admin groups!
STEP 3: deploying a VM template + change VM hardware
  • remark: no OS optimization enabled in vCenter
  • Issue: different HAL is impossible to know (i.e. single-processor/multiprocessor)
  • remark: due to lack of VAAI plugin for Storage Array, this process takes over 15-20 minutes.
STEP 4: perform guest OS & IP changes
  • if DMZ > change host file
    • x.x.x.x backupserver backupserver.local (server VLAN)
    • x.x.x.x backupserverb backupserverb.local (backup VLAN)
  • Remark: using local host-file changes makes restore unnecessary complicated. Make these changes in DNS.
STEP 5: initialize new disk(s)
  • Issue: drive letters unknown in CMDB
STEP 6: set <old school software> client details
STEP 7: reboot
STEP 8: take VM snapshot in case restore fails
THESE 8 steps take about 1 hour to complete
STEP 9: restore C:\
  • Issue: restore procedure cannot be killed by admin > no way to know real progress without involving global team.
This process takes about 30 minutes to complete
STEP 10: restore SystemState
This process takes about 30 minutes to complete
STEP 11: reboot
STEP 12: restore Data disks
The time this process takes depends on the size of the machine.
===========================================
Somehow I have a feeling this not really the most efficient way to protect your infrastructure.
Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

Ask Me A Question – button campaign

A hot topic in the last few years has been awareness around the lack of women in IT. This goes from how we give young girls dolls but boys a plastic screwdriver, to gender equality in the household, i.e., who has responsibility for the children, to how we treat women in IT on a daily basis. At the heart of the matter, I see the gender equality in STEM (Science, Technology, Engineering, Math) education as the issue which needs to be attacked at all levels of society.

In 2009, 24% of the jobs in STEM in the US were filled in by women, where in all jobs, the average was 48%. (source)

You go, girl!

I’ve said it before: I am a Heterosexual Type-A Alpha-male with ADHD. This is a recipe for disaster when it comes to treating women or just people in general. Micro-agression is probably one of the things I am guilty of the most. Quite recently I talked about a female colleague, who is a brilliant engineer, and referred to her as a ‘great girl’. I heard myself saying it, immediately corrected myself and was lucky enough my (female) partner in the conversation did not take it as an offense. After talking it over with her, I will never say it again and I will help other people understand why they shouldn’t as well. It is condescending in every possible way.

The phrase ‘girls’ night out’ or ‘boys’ night out’ is not an excuse to call a female colleague a girl! You would never refer to a man in a professional context as a boy.

Continue reading

Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

vSphere Storage Array FEEDBACK requested

I am working on a lengthy post/presentation related to UNMAP on all levels of the stack (from Guest Volume to Physical disk). During my research I have noticed that there is no way for me to find out through the VMware HCL which arrays support T10 UNMAP. Therefor I need your help. If you find 5 minutes of your time I’d love to get a few details from your array.

  • Vendor + Model + Firmware
  • vSphere Version
  • VAAI Status (screenshot 2)
  • Thin Provisioning Status (screenshot 2)
  • Delete Status (screenshot 1)

Screenshot 1:

Look up datastore device name: # esxcli storage vmfs extent list

Show VAAI details: #esxcli storage core device vaai status get -d [device]

NewImage

Screenshot 2:

Look up Thin Provisioning status: #esxcli storage core device list -d [device]

NewImage

NOTE: I am NOT interested here in Deduplication / Compression / ZeroPageReclaim so please don’t start pitching your product in the comments.

Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

Is the End of the Federation the real question?

Sometimes you get this one single thought that triggers a whole blogpost. This week someone asked me what’s to become of the EMC Federation when Joe Tucci is going on his retirement in 2015.

To give this some background; when the EMC Federation got it’s third leg, being Pivotal, next to VMware and EMC itself, there were some chairs being shifted. Once Pat Gelsinger made the move as COO of EMC to become CEO at VMware (2012), Paul Maritz moved into EMC to later become CEO of the third leg Pivotal (2013). In that light Joe Tucci said he wouldn’t leave the throne before 2015. Today we are 1 quarter away from 2015 so it’s worth thinking what could happen to the Federation.

[edit: re comment Greg Schulz] I forgot to mention that David Goulden is the 3rd CEO managing EMC Information Infrastructure (the EMC you know) and Joe Tucci is the CEO of the Federation itself.

JoeTucci

What’s to become of the federation

I think the what’s to become of the federation lies in the other question what’s to become of VMware? I think VMware is heading a very challenging time. VMware is slowly but certain losing it’s market domination. If you simplify the last 30 years in technology you could say that

  • 20 years ago everyone ran the same mainframes
  • 10 years ago everyone ran Windows physical server
  • the last 10 years everyone ran VMware hypervisors

I’m sure you’ll disagree and tell me you ran other stuff but I am speaking about a specific technology being the primary for the market ecosystem.

Continue reading

Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

VMworld wrap-up: the EVO-Family

The Announcements

It’s hard for #vExperts, bloggers, industry experts, … to be surprised at these events. We are so close to it all the other 51 weeks of the year that most announcements are either already known through some chatter on the Social Media platforms or we may already have been briefed. Another reason is that most announcements are just evolutionairy and fall in the category of ‘this was what we expected at some point‘.

EVO Rail

EVO-Family

We have been ‘warned’ that VMware was going to announce a hyperconverged platform at VMworld 2014. Under the working title Marvin we had a fair share of guessing going on. First of all I must applaud VMware for NOT going into hardware selling. VMware is and remains to the day a pure software company that simply does not have the internal structure for sales, distribution and support of hardware appliances.

This brings us to the why of EVO-family. If you simplify it, EVO (both Rail & Rack) is nothing more than VMware software, distrubuted by the OEMs. This is not new as those vendors have been selling it that way for years. Then it was only ESX(i) because that was all that was available. So why the big buzz then? In my opinion EVO is not an intent to sell more VMware. But it is a product that gives the OEM vendors an answer to the node-based datacenter architecture where Nutanix, SimpliVity and ScaleComputing have led the way.

Mike Laverick 2014 Aug 29

Continue reading

Be Social and Share: facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

The IT Hulk thoughts

%d bloggers like this: