Wednesday, May 11, 2011

EMCWorld Virtual labs rock!

I just got done performing several virtual labs here at EMCWorld and they are great. I have been hearing about all this new technology: VPlex, VBlock, VNX, UIM, and so on, but havnt had a chance to try any hands-on activities to feel the technology.
I just provisioned storage across the country, deployed enterprise level VMware services, and tiered multiple levels of storage within the span of an hour. I will be heading back this afternoon for some SRDF, FAST-VP, and Isolon labs.

My next hope is that they can do something similar to http://www.microsoft.com/events/vlabs/default.mspx so they are available all year long

How do I know I can trust your cloud?

Everyone out there is a cloud today - you have cloud apps, cloud servers, cloud services, cloud anything. This is all well and good for consumers where a couple of dollars are at risk, but what about Enterprises? When an enterprise moves to a cloud, it is a game changing and potentially critical event. If the cloud fails, people can lose jobs, customers lose trust in the enterprise, and the company could go under.

So if my company is on the line, how do I know if I can trust your cloud? Can you guarantee the availability and redundancy I require? Will you be reponsive to my needs? Will I be aware of the changes (big and small) that you make that will effect my environment? And, if you fail to meet these needs, will you compensate me for lost revenue?

A keynote at EMCWorld discussed this idea of how to establish trust with cloud provders. In the finance realm, auditors come in to review financials and controls, so why not for the cloud world? RSA released a service known as the Cloud Trust Authority (http://www.rsa.com/node.aspx?id=3861). Instead of doing your own research on who to trust and how much to trust them, RSA is creating a scorecard for cloud providers that can be shared wiht potential customers similarly to how financial audits are shared. This allows customers to compare scorecards of providers to determine at a glance who to trust, and who needs more work.

Open Source Cloud Framework

At EMCWorld I learned about the worlds first open-source cloud framework - http://cloudfoundry.com/. Developed by ex Google employees, this is a ground up Platform as a Service (Paas) environment that allows the enterprise, private, and public clouds to benefit from the same features, functions, and APIs.
Now, you can develop LOB apps on the platform and keep it secure behind your corporate firewall. Then, when the need is great enough, you can migrate it to a public cloud, or enable your private cloud available on the internet.

No longer is PaaS an all-or-nothing situation. You can integrate your internal thick apps with your internal cloud apps.

EMCWorld - 2 days down

Day 2 at EMCworld ended with a bang. Lots of great sessions and lots of fun. So far I have seen:
  • The future of consumer driven IT
    • Enterprise IT growing and being as nimble as consumer technology
  • Virtualized information world, where people no longer work on "documents" at "deskops" but work with information from anywhere
    • Imagine the portability of apps like facebook and twitter applied to LOB apps
  • Great advances in Avamar to backup and restore VMs
    • 6.0 includes block-level restoration processes
    • This could allow for an always on hot-site capable of near immediate fail-over
  • VPlex GEO
    • Enabling active/active datacenters and workload migratin across the country
  • Exchange BCDR practices and how to imporve them with RecoverPoint
  • Project Lightning - allowing FASTCache capabilities on individual servers
    • This could speed up large Databases, key LOB systems, and even VMware
    • This could also lessen the backend SAN requirements and allow lower cost disks
  • Advances in Oracle and ASM
    • Moving management of ASM volumes from the DBA to the storage admin
    • This allows the DBA to worry about the DB environment, and the storage admin to worry about the storage and performance
  • Database Cloud
Thats only the first few pages of notes! More to come soon

Tuesday, May 10, 2011

How much does your virtual environment cost?

Most companies are implementing some sort of virtualization today, but how much does that virtualed server cost? If you have 300 virtual machines, how much does that test/dev system cost? Or that font-end web server? And more importantly -- How much would it cost to move that to a public cloud?

The Accelerating the Journey to Your Cloud keynote at EMCworld today touched on this subject. How much are you paying for each VM over a 3 year period? Is that VM really that important? Should you restructure your virtual environment to be more cost-competitive with public cloud providers?

I for one am going to do some number crunching when I get back to the office to see how much we are costing, which systems/environments cost the most, and where we can streamline our environments.

Is your company ready for future storage?

Is your company ready for the future of storage? Are you?
EMCWorld session Managing information storage - Trends, Challenges, and Options 2011-2012 discussed some of the recent history of storage management and challenges, and the moved on to vew the future trends and possibilities. The whitepaper (link below) goes into great detail about storage environments, trends, and skillsets needed for tomorrows storage environments.

http://www.emc.com/collateral/emc-perspective/h2159-managing-storage-ep.pdf

Oracle Database Storage Admin Guide

In the Storage Administrators Guide to Deploying and Managing Oracle DB Clounds on EMC Storage session at EMCWorld, they mentioned the Oracle Database Storage Administrators Guide - a guide to configuring and managing Oracle disks for non-DBAs.Some quick googling came up with the below link, this contains some great information about managing ASM as the storage person, instead of expecting the DBA to know the intimate details of disks, FC, and SANs.

http://download.oracle.com/docs/cd/B28359_01/server.111/b31107/toc.htm

Oracle Database Cloud

I was sitting in the Storage Administrators Guide to Deploying and Managing Oracle DB Clounds on EMC Storage at EMCWorld, and the brought up an intersting idea - a database cloud.

I had previously heard about Oracle Real Application Clusters (RAC), a clustered, scalable, high-performance database solution. I had also thought about putting multiple Oracle databases into a RAC cluster (small, medium, and large DBs) to save on HW costs. But I had never thought about a database cloud before.

Microsoft has their Azure DB platform - a cloud based SQL db, but it is hosted in the public cloud and you have to customize programs to use it. What if we could do something similar in our private cloud to make DB hosting as easy as VM hosting is in VMware?
  • New application coming onboard? Just submit a webform and the DB is spun up.
  • Existing application needing more resources? Just click a button in the UI and more memory and disk is provisioned.
  • Need high performance at night to calculate reports, but low performace during the day? The cloud over-subscribes resources and moves them as needed.
The only problem I see in this is that some applications are tied lock-step to the DB release - especially Oracle products. This could be a problem with multi-tenancy
  • Want to upgrade your app? Then upgrade your DB as well.
  • Want to fix a bug? The install a patch on your DB

New features in Avamar Desktop/Laptop

Avamar 6.0 just went RA (apparently 1 step prior to GA) and the EMCWorld session EMC Avamar: Edge for Remote Office and Desktop/Laptop Data Protection, they discussed some of the new features.
They started off the session discussing the features and capabilties of the Desktop/Laptop backup solution. Basically it is the same as the serve backup solution, tweaked for end-user friendliness.
They then discussed how EMC IT rolled out Avamar to their desktop users. At EMC they chose to roll out an entirely separate grid infrastructure for desktops, but mostly due to the number of clients and unique timing requirements. They already had a large Avamar grid environment for server backup that was tuned and scheduled for server environments. Of main importance was the fact that server backups occur at night, while desktop/laptop backups would happen during the day (when the computers are turned on).
To handle the timing and maintenance requirements of desktops, as well as the thousands of clients, they implemented a new grid environment throughout the world.

New in the 6.0 Desktop/Laptop solution are end-user improvements. The UI that users have is much more complete and easy to use. Additionally, in 6.0 they now allow a user to restore data from a profile on a different machine. In 5.0, they only allow restores by the same user on the same computer, but if you use multiple computers, upgrade, or replace your system, an Avamar admin had to assist in the restores.

Coming up in future releases (no dates or confirmation on release)
  • Native encryption across the internet
    • No longer will our CEO fly across the country without being backed up. We could configure his laptop to backup across the internet, and then when at other companies or at the hotel, his system will backup across the internet without him connecting a VPN.
  • Keep last backup
    • If you have a short retenion time (less than 30 days), it is possible that a system could be offline long enough for the backups to be purged. If the system crashes after the purge, then all system data is lost
    • Scenario: Someone takes their laptop home for maternity leave. They occasionally use OWA to send email, but never connect to the VPN to backup. 30 days into the leave, the laptop falls to the ground and is destroyed.
    • There is discussion of an option to keep the last backup of a system even after the expiration date. This could be selected on individual systems that IT is alerted to people being offline for an extended time.

Monday, May 09, 2011

Storage Lightning

Solid State Disks are revolutionizing the computer industry. It started out as a method to quickly store and retrieve data, but EMC has taken it to the next level with its FAST and FASTCache features.
Several vendors have released PCI cards that are plugged into a server and serve as SSD disks, this disk is then used for super fast drives in databases. EMC announced today that they are entering the arena with a project nicknamed Lighting!
Instead of simply presenting another disk, EMC is positioning the Lightning product as an integration between local disk, SAN disk, and FAST. This means that instead of having to manually identify what data should be local, and what should remain on the SAN, the system will handle the migration automatically.

The specifics are still a little light (its only in testing phase now), but I can see great benefit of this. Not only in high-performance databases, but also in VMware workloads where expensive disk is required to handle the workloads and boot storms can cripple an environment

More detail can be found at the press release: http://www.emc.com/about/news/press/2011/20110509-05.htm

Site Resiliance

This may be old hat to some people, but I just came across my first definition of site resiliance, and the types of sites available
  • Production datacenter: This is where production workloads run normally. This can be a single datacenter, or multiple datacenters that can operate as a DR site for the others
  • Hot datacenter: This is a live environment which is pre-staged with the hardware and software configurations to handle a DR event. An example of this would be VMware's site-failover features that can turn off low priority workloads in the DR site, and turn on high priority workloads. The main benefit here is that DR failover can be automatic, minimizing the RPO and RTO.
  • Warm datacenter: This is a smaller site dedicated to prividing DR services for select business critical functions. This is normally a federated model where multiple customers contract for resources.
  • Cold datacenter: This is an offline environment that is capable of providing DR services. Often times this consists of a contract from hardware suppliers to drop-ship emergency equipment to a remote location, which then must be used to rebuild and recover the production datacenter.
While this may not be the 100% answer, it atleast puts some industry standards behind the "DR Facility" question.

More definitions are avaialble at: http://en.wikipedia.org/wiki/Backup_site.

Migrating live VMs across multiple states?

Last year at EMCWorld, they unveiled the EMC VPlex - a way of having active-active datacenters and being able to synchronize data between them in such a way that you could run virtual workloads between them. The even demo'd migrating between SANs that were theoretically in different datacenters. This was a great idea, but my DR datacenter is several states away, well beyond the required latency parameters.
Today they announced a beta customer using the VPlex GEO, enabling them to run datacenters in Pittsburg and Dallas, and still use the VPlex functionality. This could allow us to put small datacenters, geographically dispersed, and capable of migrating data and workloads between them. This could allow a 24x7 line of business application to follow-the-sun, without major downtime or customization.

EMCWorld - First session down!

Think about Facebook – you can use it on Windows, Mac, Linux, Blackberry, iPhone, Droid, and any other web enabled device. This is true application as a service, wherever you are, whatever platform you are using, you can use Facebook. On many of these devices, you can install an app to use this software even easier.Now imagine the same thing in your corporate environment. My company has Windows, Blackberry, iPhone, iPad and even a couple of Macs, way too complicated for a traditional Windows company. This is exactly what VMware End User Computing Vision 2015 – Virtual Workspaces in the Cloud targeted.
Instead of managing devices, we should be able to manage the users and the user data. By virtualizing the apps through different methods, the device being used shouldn’t matter. The end user can use their Office Suite or LOB app regardless of being at work, home, airport, or whichever device they want.
Seeing how the explosion of the app-store has taken over deployment of consumer applications, I can definitely see something similar for the corporate environment.

Just became and EMC Proven Professional!

I saw that I could sign up at EMCWorld to take the EMC certification exams for half price, so I thought it migh be a good idea. Every year they have asked me if I wanted a cool shirt, and when I said yes, they said I had to pass the test -- this year I did.

I dont know what I can say about the exam, but I would mention that it was more difficult than I thought. The breadth of technologies in question (Celerra, Symmetrix, Clariion, FC, iSCSI, etc...) made it very difficult. This is probably the first time where I feel a 60% passing grade is acceptable.

Now, on to the Cloud Architect certification!

Wednesday, May 04, 2011

Failover mode for VMware on CX4

I recently realized that our default VMware environment using our EMC CX4 may not be using the ideal pathing configuration. Our environment consists of ESX servers with 2 HBAs connected to 2 FC switches. Each FC switch is then connected to both Storage Processors on the EMC SAN. This means there are 4 paths from the ESX host to any given LUN, but VMware only actively used 1 path by default.
You can confirm this by selecting a host and clicking the Configuration tab and selecting Storage. Select a VMFS datastore and click Properties | Manage Paths. If your environment is like mine, Path Selection will be Most Recently Used and only 1 path will be listed as Active (I/O).

To improve this, 2 things need to happen: The Path Selection policy in VMware needs to be changed, and the FailOver mode on the SAN needs to be changed. To change the Path Selection policy, click the drop-down list and select Round Robin and click Change.
To change the FailOver mode we can either use the GUI or command line. In the GUI, select Tools | Failover Setup Wizard and follow the steps, choosing failover mode of 4. To use the command line (useful if you have more than a handful of systems) do the following:

Find current failover mode:
NaviSECCli.exe -user username -password password -scope 0 -h sanAddress port -list -failovermode
This will list the failover mode for all hosts connected to the SAN. Once you find the host your interested in, you will see the Failover mode: is set to 1.
Change failover mode:
NaviSECCli.exe -user username -password password -scope 0 -h sanAddress storagegroup -sethost -host hostname -failovermode 4

Once the failover mode is changed, reboot the ESX host and then review the path configuration for the LUNs. If all went well, the Path Selection should be Round Robin, and Storage array Type should be VMW_SATP_ALUA_CX

Monday, May 02, 2011

Extracting music from YouTube

DISCLAIMER: I am not fully aware of the legality of this. I am assuming that it is OK since I am not doing any hacking or breaking of cryptography, but your mileage may vary.

Below are the high-level steps I used to download music from YouTube and then burn to a CD. I performed these steps on Ubuntu 10.4
  1. Create a folder to hold the video and audio files
  2. Find the music you wish to download and copy the URLs to a temp file
    1. Such as http://www.youtube.com/watch?v=t_bLwrcdJgA
  3. Using get_flash_videos, download the files
    1. get_flash_videos http://www.youtube.com/watch?v=t_bLwrcdJgA
  4. Use mplayer to extract the WAV data
    1. for i in `ls *.mp4`; do mplayer $i -ao pcm:file=$i.wav; done
  5. Use normalize-audio to make all the music the same volume
    1. normalize-audio -m -v  *.wav
  6. Finally, use your favorite CD burning software to create the disk.
You can research the commands individually to get more information on what they are and how they work.