"...their vision has been instrumental in our adoption of these technologies before they become buzzwords." - Network Manager at Nelson Mullins
"I would simply say that they do the work you need and do it very well." - Director of IT at Piedmont Healthcare
"Not only are the eGroup engineers extremely good, they have all been able to fit in with my existing staff and build relationships that are beneficial for all of our projects." - VP and CIO, Beaufort Memorial Hospital
We called around to other companies to see who they used and over and over the name eGroup was given. After the first meeting, I knew they had the knowledge and staff to meet our needs." - IT Director, MCAngus, Goudelocke and Courie
"I was skeptical about the amount of work they wanted to do in a short amount of time. But we were able to get it all accomplished in less time than they expected. Totally blew us away." - IT Manager, Horry Electric Cooperative
"With their expertise at our disposal, it is like having additional staff on hand. They have never made us feel like we were just another client. More like colleagues." - IT Manager, Horry Electric Cooperative
Having put in several Avamar solutions in the past few months, I wanted to share some comments and statistics from real customer environments that are now happily globally (and source) de-duplicating their backups with EMC Avamar.
Below are comments from our most recent Avamar victory.
“At this point I have calculated an approximate 72% de-duplication rate for all servers. Our backup windows for daily backups has shrunk from 8 hours to 1.5 hours. Our weekend backup has shrunk from 48 hours to 1.5 hours. ”
48 hours to 1.5 hours. That’s kind of a big deal. Going from almost NOT getting the backups completed over the weekend, to getting them done before kickoff of the early Saturday game. Makes for a relaxing football season.
Below is a screenshot from a different customer environment running Avamar. I cut off their client names to protect the innocent. This customer has dozens of servers that have a de-duplication ratio of OVER 95% (with 10+ that are 99%).
The “on demand backup” jobs in the screenshot were first time backups that had the benefit of global de-duplication. Only a small percentage of their data was NOT already on the Avamar grid from another client, so Avamar did not need to send that duplicate data (that’s the GLOBAL de-dupe action).
A good example to illustrate what this means: Consider the Operating System files in the following scenario. Client A runs Server 2008 R2 and has already been backed up. Client B also runs Server 2008 R2. Same OS.
The initial backup for Client B will NOT send any duplicate OS files over the network to be backed up, because the data is already on the Avamar grid (that’s what duplicate means, right?). In the event that a restore is needed, Avamar knows how to get that data out and back to the client. It’s like magic. And it’s all done with variable length data segments; not at an entire file level, and not with a fixed block size.
Check out that 971.5GB backup that completed (and REALLY completed, not that “Completed with exceptions” nonsense) in 2 hours. That’s almost 500GB an hour.
Ah, the power of Avamar. Can your backup system do that?
As we design Business Continuity solutions for our customers; one of the biggest challenges in the past has been addressing the siloed nature of their independent data centers with respect to networking.
Consider this: Your primary data center has one IP schema and your DR data center has another. If we were to setup an Active / Active hot site scenario where you could move your compute loads between them, we would have to change the IP address of each of one of your systems as they moved through your cloud infrastructure. The automation of moving the actual compute load is not a difficult problem to solve with tools like VMware’s Site Recovery Manager.
In the past to have the IP addresses change automatically it took a lot of manual scripting and fair amount of skill with XML. I know what your thinking, “It would be really nice to encapsulate the virtual network layer in this scenario”….hold that thought.
A little while back at EMC world 2010 – VMotioning over long distance (WAN) was displayed in a truly geo-stretched environment using EMC’s VPLEX. The speed at which these VMotions where shown over a WAN were staggering. However, the first thought that popped in my head was; that’s great, but what happens when the VM that was sent from Chicago to New York can’t communicate on the network because it has a different IP schema.
Enter the VXLAN, a truly scalable way to overlay the network (layer 2) for Cloud Computing models.
VXLAN – Virtual Extendable LAN has the following features.
• Logical networks to be extended among virtual machines placed in different subnets
• Flexible, scalable cloud architecture in which new servers can be added in different subnets
• Migration of virtual machines between servers in different subnets
Designing Business Continuity solutions just got easier.
More to come soon as we test the VXLAN functionality in the eGroup Lab and the NEXUS 1000v 1.5 beta.
There are times when an Exchange Mailbox Database is unavailable – wether it is missing or the underlying storage is unavailable. Recently, while visiting a customer, they asked about the best method for solving this issue when the old database does not need to be retrieved. There is a simple set of commands that can be used to move the path and generate a new, empty, .edb file.
To identify the the database to work with one can either use the Mailbox name or the Mailbox GUID. To obtain the GUID run the following Exchange Management Shell command:
The -ConfigurationOnly property instructs the Exchange Management Shell to ignore the fact that it cannot find the current path or file. The above command without the -ConfigurationOnly property is identical to running the Move Database Path option within the Exchange Management GUI.
Once the above command is complete, the Mailbox database can be mounted at which time it will create a new, empty, .edb file in the new storage location.
This is a nice command to have in your brain bank when faced with a situation where the existing path for your mailbox stores cannot be contacted or you’re restoring Mailbox databases to a new location.
Someone on Twitter the other day quoted Paul Maritz, CEO of VMware, as stating that “at any given time, there are more VMs ‘in flight’ than there are aircraft”. That describes just how heavily vMotion is used in VMware environments.
It’s the most used feature in VMware environments by a LARGE margin, and for good reason. And now with the release of vSphere 5, VMware has made our favorite feature even better.
First, vMotion can now use multiple network adapters to increase the speed in which VMs are migrated. The vmkernel transparently balances the load across available vMotion vmkernels (and their physical NICs)– even if you’re only performing a single vMotion. VMware’s internal testing showed migration times improved by a factor of 3x.
The best practice around this dictates that you should use a single vSwitch, and create a vmkernel for each physical NIC you want to use– and then go into each vmkernel and set a single NIC as the Active NIC . I showed this in a previous video (see below) and the process is the same (skip a few minutes in to see the “how to” within vSphere).
Also helping with the speed of migrations, and helping to reduce failed vMotions, is a new “Stun During Page Send” (SDPS) feature. What this feature essentially does is temporarily slow down the speed in which a VM modifies its pages in memory to allow a vMotion task to copy the contents of memory (iteratively until the “final” copy) to the destination host. In vSphere 4.1, if a VM is modifying its memory faster than the contents of that memory can be copied over the network, the vMotion stalls and eventually fails. SDPS is designed for this specific situation.
In addition, the performance in which memory traces are performed (to track the modified pages in memory during a vMotion) has increased. This reduces the impact / overhead on the virtual machine’s workload (and application performance) during a vMotion.
Also in vSphere 5, “Metro vMotion” supports vMotions across networks with up to 10ms of round trip latency, as compared to up to 5 ms of round trip latency in vSphere 4.1. This extends the capability to move machines between geographically disparate data centers (though 10ms of latency usually indicates the sites are still pretty close to each other).
The current version of EMC’s Unisphere for the VNX does not allow for more than one storage array to be present in a single “management domain”– meaning you cannot (today) manage multiple EMC arrays from a single Unisphere session on the VNX. Bummer. While this has been promised for a future release, it’s a known limitation today.
And generally speaking, it’s not a major issue, but rather a temporary inconvenience.
However, having more than one array in the storage or management domain has traditionally been a requirement of EMC’s SAN Copy and MirrorView array-based software solutions. And when migrating data from a prior generation EMC array (or even a non-EMC array) to the VNX, these tools (SAN Copy specifically) come in very handy.
The good news is that there are documented workarounds to use SAN Copy and MirrorView on the VNX. Here is the video for using the LUN’s “WWN” unique ID to create a SAN Copy session between a CX3-40 and a VNX 5500. I’ll post a separate video for MirrorView in a few days.
Also, make sure you know the ports that your array uses for SAN Copy / MirrorView, and (assuming a FC network) zone them appropriately. For example, on the VNX arrays used in the videos above, SP port “0″ are used for MirrorView, and SP port “1″ are used for SAN Copy. They CANNOT share the same SP port.
The SharePoint August 2011 Cumulative Update has been released, but has not been officially announced yet. For those of you that are having issues with the User Profile Service, this update may help.
With this update, Microsoft is changing how it deploys updates for all Office Server products. There is now ONE update file for all languages and SharePoint products. This means that if you are running WSS 3.0, MOSS 2007, SharePoint Foundation, SharePoint 2010 Server, or Project…you just need to download one file. Much more convenient!
Please note: You must have Service Pack 1 installed before installing this update.
You can get more information on this release and links to the downloads on Stefan Gobner’s blog here.
Here is a list of fixes from the SharePoint 2010 Server KB2553048 page…
The Activity Feed displays incorrect colleague information in SharePoint Server 2010. This issue specifically occurs when the colleagues are added through the User Profile Change Job in the User Profile Application.
Assume that you plan to configure a user profile connection in SharePoint Server 2010. However, you cannot add a new mapping for case-sensitive string type attributes.
Assume that you plan to configure a user profile connection in SharePoint Server 2010. However, you cannot add a new mapping for print case string type attributes.
Consider the following scenario:
An Active Directory group has a name that is longer than 255 characters.
You create an AD synchronization connection to the AD group in SharePoint Server 2010.
You perform a full synchronization.
In this scenario, the full synchronization does not finish successfully.
Consider the following scenario:
A user posts some comments on a Note Board Web Part in SharePoint Server 2010.
You delete the user profile of the user.
You view the comments on the Note Board Web Part.
In this scenario, the comments that the user posted are invisible.
Assume that a custom Activity Directory attribute is a numeric string type attribute. In this situation, the Active Directory attribute cannot be mapped as a new property in SharePoint Server 2010.
You restart the User Profile Synchronization service in SharePoint Server 2010 after you apply the hotfix package that is described in Knowledge Base article 2536599. In this situation, an exception occurs.
In SharePoint Server 2010, the Audience Picker tries to reload a selected security group that contains a space character in its name. However, the Audience Picker cannot resolve the security group name, and it leaves the security group as an unresolved entity.
Force Protection Industries is a global leader in manufacturing heavy armored vehicles for the DOD and other military organizations. Force Protection currently operates a 400,000+ sq. ft. plant with a workforce of more than 1,000.
Challenge:
After experiencing serious email outages and turnover in support staff, Force Protection was looking for a partner who could help improve email stability in order to ensure delivery of quality service to their end user.
Solution:
eGroup provided hardware platform upgrades within Force Protection’s network and applications as well as system integration support services.
Results:
After partnering with eGroup, Force Protection has experienced a performance system increase of 25% and delivery of stable services increased 75%. In addition, eGroup has provided Force Protection with a stable base to build upon for future upgrades as the organization continues to grow.
“[eGroup takes a] methodical approach to assessing problems or challenges and providing a plan for remediation.” – Network Infrastructure Manager at Force Protection Industries
