In a previous post, we discussed the incredible WAN performance gains made possible by combining Pancetera with a Riverbed Steelhead appliance. Given that networks and backups are often discussed in different terms, let’s look at what the technology really does accelerating VM backups over the WAN. Today, backup managers face the daunting task of backing up systems that operate 24×7 in an ever-shrinking backup window of relatively idle use (typically midnight to 6 AM). Yet, critical items, such as virtual machines, cannot be backed up without interrupting operations. Also, the need to protect against a regional disaster means that data must be replicated off-site, typically over an IP-based WAN. There are quite a few challenges involved in this scenario including backup and replication windows, reliability of the system, and cost.
As described earlier, combining Pancetera and a Riverbed Steelhead appliance reduces the time to replicate a VM by 95% This means that backup managers now have a potent solution for working within their backup / replication windows. More importantly, this solution can actually SAVE an organization $500,000 per year. For example, a dedicated OC-3 line of 155 Mbps with Pancetera and Riverbed Steelhead will perform better than an OC-12 of 622 Mbps line at a quarter of the cost. Savings of $500,000+ per year provide a near instant ROI.
Other benefits include: backing up running VMs, drop-in installation without having to customize the backup system, streamlining redundant or end-of-life components such as VMware Consolidated Backup (VCB), and more.
Our web page on the solution can be found here.
Tags: acceleration, backup, cloud, migration, replication, roi, wan
Posted September 4th, 2010 by Mitch in cloud, data protection, virtualization | No Comments »
We are very excited about VMworld 2010 in San Francisco at the Moscone Center. We’re exhibiting in the New Innovator Pod 1319D and invite everyone to swing by and check out a demo of Pancetera Unite. Some of our sales and marketing folks will be at the booth and our founders will be around from time to time. We have some great giveaways at the booth as well–totally awesome luggage tags with LEDs that blink when your bag comes off the plane for your trip back home!
We are also sponsoring the VMworld Hall Crawl on Tuesday at 4 PM. Come by and join us for a beer! We look forward to seeing you at VMworld!
Tags: events, San Francisco, vmworld
Posted August 29th, 2010 by Mitch in events | No Comments »
We hear a lot of excitement about APIs these days in the world of computing. APIs let third parties extend systems beyond what the creators of those systems originally imagined. Look at all the applications that have been built on Amazon’s Web Services or Twitter’s APIs. In the last few months, there’s been a lot of chatter about various APIs related to virtualization, in particular the vStorage APIs from VMware.
The interesting part of this is that the ink and excitement about the VMware vStorage APIs has been in the press and conversations that are not developer audiences. Why should end-users get so excited about particular APIs? There’s nothing there for the end-user to “touch”! Clearly there has been a lot of excitement about what vendors will bring to market with the vStorage APIs.
Often we are asked, “How is it that you support so many applications with Pancetera? Did you have to create plug-ins for all these applications for integration with Pancetera Unite?” The beauty of our approach is that we transform the run-time and storage complexity of a virtualization environment into a file system structure that any tool familiar with file systems can navigate. All the normal operations you’d expect are there, such as chdir(), open(), read(), and so on. This means existing enterprise tools from backup to replication to a drag-and-drop copy with normal desktop file tools all work with Pancetera Unite out of the box. No plug-ins required. This has let us ship Pancetera Unite with support for dozens of world-class enterprise tools out of the box with no integration or special coding required. Our customers are reporting successful deployments of Pancetera Unite with tools we haven’t tried ourselves and things just work.
As the virtualization marketplace evolves, we expect to see more diversity in vendors and environments. We’ve designed Pancetera Unite to enable us to add support for more virtualization back-ends, such as Hyper-V, while preserving the file system that we present on top of virtual storage complexity. This means customers can mix best of breed technologies for their specific use cases without concerning themselves as to which vendors on the front-end support which APIs out the back end. Want to use your search tools with VMware virtual machines? We’ve got you covered. Want to use rsync with Hyper-V? No problem.
In fact, a lot of work has been done around file system optimization and aggregation by a number of vendors, whether they are building file system virtualization products or WAN accelerator products. In our tests, we integrate well with a wide variety of these products, enabling customers to maximize value and re-use existing infrastructure. We know organizations do not have the time or money to throw away perfectly good solutions that are in place today—and we let organizations continue using those existing deployments while integrating them with greater efficiency into their virtual infrastructure.
Tags: api, file system, vstorage
Posted August 20th, 2010 by Mitch in integration | No Comments »
Typically when we’ve discussed the need to acceleration of moving VMs, we’ve talked about moving VMs across the WAN. However, VM migration and movement challenges occur on the LAN as well. Sure, standard tools exist for moving VMs around within datastores that have been associated together in a vCenter cluster, but what about moving test VMs into production? Do you organize test as part of the production vCenter or do you have to bring in separate tools? If you do organize test as part of the production cluster groups, how do you avoid test VMs going into production without relocating them into the production cluster? I’ve heard from customers on several occasions that despite their process for migrating from test to production, sometimes VMs in test become production.
“VM Mobility” refers to the ability to move or migrate VMs. We’ve started using the term to mean a movement of VMs more broad than a simple vMotionTM, but also other operations, whether going from test to production, cloud bursting, moving a VDI instance from a repository to a client workstation. Over time, VM mobility will come to mean not only moving between homogeneous hypervisor environments, but more interesting mobility scenarios—from a Xen cloud service VM to a VMware Workstation VM on a laptop. Or moving a test VM on HyperV to vSphere in production. Moving VMs efficiently and quickly continues to stress even the best IT infrastructure designs. We think there lots of room for making this a much, much easier process. Shouldn’t moving VMs from one datastore to another or even between hypervisors ultimately be as easy as moving files between directories?
Tags: cloud, migration, mobility, storage, vmotion, vmware
Posted July 31st, 2010 by Mitch in cloud, virtualization | No Comments »
Pancetera has been running a survey looking into Key Virtualization Challenges to get a better understanding of the issues our customers face in their virtualization environments. The data has proven to be illuminating. You can take a look at the results here. The key takeaway is that scaling virtual environments is proving to be a difficult task.
Key Challenges
Storage Complexity and Cost were your primary pain points. These were followed closely by Backup Complexity and Network Performance. In speaking with customers, we are finding that virtualization scales to a point in most environments when progress slows down as you search for solutions to the challenges that impede further growth.
Server Virtualization Footprint
VMware continues to have a dominant market share in server virtualization, but Microsoft Hyper-V and Citrix XenServer are making significant inroads. They were each present in 17% of your environments. This compares with VMware ESX at 77% and VMware ESXi at 40%. No other hypervisors are in the running. (The percentages add up to more than 100% because some of you are running more than one type of hypervisor.)
VCB
VMware Consolidated Backup is still alive and kicking in 21% of your environments, even though VMware has announced its end of life. Pancetera provides a drop in replacement for VCB that works with any existing backup agents and enables you to re-purpose the servers and storage currently dedicated to VCB, so if this is of interest please register for a 30 day free trial.
Tags: challenges, hyperv, survey, vmware, xen
Posted June 28th, 2010 by Mitch in virtualization | No Comments »
In a previous post, I wrote about our virtualization survey indicating strong interest in moving virtual machines across the WAN. Recently, our engineering team has been doing some experiments with enterprise WAN accelerators to see how using Pancetera in conjunction with these products can benefit customers.
The results are absolutely incredible. Leveraging the combination of the Pancetera solution with enterprise WAN accelerators enables customers to increase VM replication speeds by 10x, or over 6x the speed that a WAN accelerator offers alone. The graph below shows the results for moving an 8 GB VM with Windows 2003 installed over a 10 Mbps pipe. A naive move with no extra tools requires a little over 2 hours to move the VM. Adding a WAN accelerator product on both ends of the pipe without Pancetera results in a respectable improvement—the VM is moved in 1 hour and 20 minutes. Adding Pancetera to the path with the WAN accelerators lowers the time required to just 13 minutes.
How many VMs could you move on your pipe? If you could move your VMs 6x—10x faster than you can today, how would that change your virtualization environment? This massive acceleration in the movement of VMs over long distances is going to enable of lot of benefits for our customers. Perhaps “cloud bursting”—moving VMs dynamically in and out of the “cloud” as performance requirements dictate—can suddenly become practical? Try us out and let us know what you think of the benefits of Pancetera SmartRead in your environment.
Tags: acceleration, cloud, migration, replication, wan
Posted June 25th, 2010 by Mitch in cloud, virtualization | No Comments »
There’s a lot of excitement around all things “cloud” these days. Folks are talking about moving virtual machines between the data center to the cloud or through the cloud to another site. There’s interest in pushing storage to the cloud, services to the cloud, and so on. At Pancetera, we use a number of cloud services for email, calendaring, collaboration, and our phone system.
Recently I’ve been talking to customers about having the ability to push configuration data from our virtual appliance to ease stamping out new appliances at other sites or in different infrastructure without manually reconfiguring a set of base parameters. The idea is similar to what I’ve seen in a number of consumer applications lately, where we recognize that customers often want to use the same software with the same configuration parameters on a number of computers in different geographies. I’m surprised that I haven’t seen more discussion and interest about pushing configuration up to the cloud, whether as a mechanism to facilitate simple re-instantiation of that configuration or as a backup of that configuration. As people move virtual machines around, move services around, and mix local applications with remote back-ends, centralizing the management of configuration is going to grow in importance. I expect to see more vendors that are shipping enterprise software that operates in the data center to add more cloud management and tools that extend the non-data paths of the data center out to the cloud.
Tags: cloud, management, saas, service
Posted June 22nd, 2010 by Mitch in cloud | No Comments »
We recently ran a survey asking folks what their biggest challenges are in managing their virtualization infrastructure. We were surprised to see that 1 in 5 of the respondents are facing significant problems related to moving virtual machines over the WAN. We have spent time drilling into this issue with customers to understand how people are moving virtual machines today. We found three clear use cases: One is long-distance storage VMotion of a running virtual machine; another is off-site replication for a stand-by or backup copy, and a third is to move a virtual machine so that another instance of that server can be instantiated at another location, such as a remote office, vault, or a cloud provider.
The long-distance VMotion is clearly a vCenter function and can be assisted today by WAN accelerator products in the WAN data path. Off-site replication or moving a virtual machine can be coordinated with various tools, but today those tools need to be aware of the virtual infrastructure. Customers can buy specific point solutions around this use case or build their own scripts to assist, neither of which is particularly appealing.
With Pancetera, customers can use standard file system replication tools and techniques to move running virtual machines to new targets, without disturbing the running virtual machine, whether on the LAN or across the WAN. We enable this use case without scripting. With the Pancetera SmartRead technology, we can also further accelerate the speeds that are achieved with WAN accelerators that customers have today.
If you are having a virtual machine migration problem, let us know. We’d be happy to help.
Tags: cloud, deduplication, replication, wan
Posted June 3rd, 2010 by Mitch in cloud, virtualization | No Comments »
Recently I saw a thread on a tech news site about a major cloud vendor losing customer data. In the cloud vendor’s support forums, some folks were posting that the customer should have been leveraging the vendor’s snapshot feature.
Seeing this discussion reminded me of similar conversations we’ve had when discussing backup. We are sometimes asked, “The hypervisor vendors support snapshots. So why do I need to backup my virtual machines?”
Snapshots are a handy technology but they are not a backup solution. When snapshots are used as a backup solution, you are inviting disaster (data loss). There are many facets to a proper backup solution–a set of known restore points, a method to create additional copies of the data on a second set of storage media, a way to move that media to a separate physical location, a process to retrieve those bits from the off-site location and re-instantiate those bits into an operating environment. There are often other elements to the backup process, such as compliance or retention policies.
Out of these basic criteria for backup, what do snapshots by themselves offer? As it turns out, snapshots only provide a set of restore points and a way to “go back”–if you remembered to take a snapshot. Snapshots by themselves often do not create a second set of data, and certainly not on a second set of storage. “Ah-ha, but if we replicate the snapshots, then we have backup!” Sure, replicating snapshots enables more of a solution but can you orchestrate this process, apply it consistently across your organization, test recovery, and archive the data, without bringing in traditional backup software?
Backup isn’t sexy, but it continues to be a critical part of production IT infrastructure. Trying to cut corners by substituting snapshots in place of backup invites serious pain down the road. If you’re using cloud storage, ask your provider what they offer beyond snapshots for data protection. Are they using enterprise backup software to protect your data? Can you integrate your enterprise backups with these cloud storage pools in an efficient manner?
Today we find customers are struggling to backup their virtual machines with their traditional enterprise backup products, and we’re here to help.
Tags: backup, cloud, enterprise, replication, snapshot, virtualization
Posted May 10th, 2010 by Mitch in data protection | No Comments »
We often talk to customers who are using multiple backup products to protect their virtualization environment. They are often using low-end VMware-only point solutions that are targeted for virtualization environments to backup VMDK images and standard enterprise backup products for backing up the files inside their virtual machines. While this covers their bases for both file-level recovery and image-level disaster recovery, customers find managing two (or more) solutions hugely challenging. Some customers find their backups are taking longer than 24 hours, which presents significant challenges to production operations.
At Pancetera, we see this as a stop-gap and sorely lacking as a long-term solution. Customers should be able to leverage enterprise applications in their virtualization environments without biting off tremendous I/O penalties or significant agent management complexity. Deploying enterprise applications in the data protection realm often involves numerous moving parts and current point solutions for virtualization do not address these issues. The pain of managing huge numbers of agents is so severe that some customers are knowingly not backing up significant percentages of their virtual machine environment and accepting the risk of this practice.
If this sounds like a challenge you are facing, we’d love help. Pancetera enables you to bridge your traditional agents with your virtualization environment in a manner that is efficient, low risk, and preserves your existing enterprise software and infrastructure investments.
Tags: complexity, enterprise, management
Posted April 26th, 2010 by Mitch in virtualization | No Comments »
