Server virtualisation has brought a huge boost in utilisation levels to Polish datacentres. Indeed, that’s one of the technology’s key benefits: putting many application servers on one physical device makes much better use of resources.
But such levels of concentration also put many eggs in one basket. Where the failure of one physical server usually results in a lack of availability for a single application, the failure of a host to many virtualised servers multiplies the potential challenges.
And that’s one major reason why virtual server backup is such a critical process.
Virtualised environments are commonplace in Polish datacentres. IDC estimates that nearly a third of new x86 servers sold in Poland will host virtual servers.
“Based on numerous interviews with users, we assume that all the major companies and public institutions in Poland use virtual machines,” says Jarosław Smulski, research manager at IDC Poland. “But, Polish IT administrators quickly learned that the security of virtual servers is much more important than that of physical machines.”
So while Polish organisations gain many benefits from virtualisation, they risk losing data as a result of virtual server backup provision that does not meet the needs of their new infrastructure.
More on virtual server backup
“Many datacentres use outdated systems for backup. This applies in particular to those run in-house by companies,” says Tomasz Gaj, managing partner at Hightech Rescue in Warsaw, a lifesaving technologies company.
“As a result, virtual machine backup processes do not meet expectations and administrators have reason to complain,” he adds.
Virtual server backup is different
“Currently, businesses have access to relatively inexpensive virtual servers, but administrators often ignore the differences between physical and virtual infrastructure,” says Krzysztof Madejski, deputy director of IT at Agora, one of the biggest media groups in Poland.
“But the treatment of virtual servers as if they were physical machines is a mistake. Server virtualisation can help to protect data and applications from failures, but only if it is properly implemented. There are hidden problems with virtual machines backups.”
Modern datacentres are equipped with new tools for virtualisation and create copies of virtual machines with greater competence. They can also relatively quickly restore them in a case of failure.
Artur Matzka, director of infrastructure development at IT services provider Integrated Solutions, says: “Virtual machines are not indifferent to the backup systems used. Modern datacentre environments require backup tools with more specialisation than previously. On one hand they should meet the criteria of technology, allowing for efficient management of the entire IT environment, and on the other to answer individual business requirements related to information security.”
Other important features of contemporary backup for virtual machines include data deduplication. Here, data reduction allows for shorter backup times, lower capacity requirements and faster transfer between geographically distant locations.
Another feature of new backup systems, adds Matzka, is that businesses can centrally manage system components deployed in a company’s different territorial and organisational units.
Wojciech Sobczak, IT specialist at datacentre service provider Comarch Data Center, says: ”Most applications that enterprises store in our datacentre, including business-critical applications, are supported by the virtual servers. Every day we make differential backups with volume of several terabytes and at weekends comprehensive virtual machine backups measured in hundreds of terabytes.”
Case study: Raiffeisen Bank Poland
“We could not rely on the backup,” recalls Sławomir Wolak, technical lead/infrastructure architect at Raiffeisen Bank Poland.
The bank’s IT department oversees two datacentres and operates 100 locations throughout Poland. More than 3,000 staff use applications that support the bank’s daily work. Its virtual environment consists of 34 VMware ESX hosts and 260 virtual machines that drive the critical applications.
We can identify problems with backup and restore a virtual machine in a few minutes without any interruption to service
Sławomir Wolak, technical lead, Raiffeisen Bank
“Tivoli Storage Manager was complicated and not very reliable and we did not have a proper insight into backup sets,” says Wolak.
“We implemented Veeam Backup & Replication and that allows us to carry out backups twice as fast as in the previous system. Also, the time to fully restore virtual machines is about half what it was previously.
“The implementation of Veeam has led us to lower costs and reduce the complexity of IT resources in the bank.”
Wolak’s team can manage the entire virtual infrastructure from one integrated console in Veeam, while instant virtual machine recovery enables the bank to quickly restore all virtual machines. It is also possible to run each machine directly from a compressed and deduplicated backup file.
“Currently, we are able to identify problems with backup and restore a virtual machine in a few minutes without any interruption to service,” says Wolak.
The new system provides continuous data protection, and allows the bank to record changes and update virtual machine images in short timescales for the needs of local and external replication.
“We can easily view all backup sessions and record information about the backups created in notes to virtual machines,” says Wolak.
Case study: Ergomedia
“VMware helped us to guarantee the availability and stability of the entire IT environment,” says Przemysław Skowroński, managing director at Lodz-based Ergomedia, a company focused on unified communications, outsourcing and hosting of servers and applications.
Ergomedia was the first company in Poland to host the complete package of Microsoft applications (Exchange 2010, SharePoint 2010, Lync 2010 and CRM 2011) for small and medium enterprises.
Unsurprisingly, therefore, the core element of its virtualisation strategy was Microsoft Hyper–V.
But the company became increasingly dissatisfied with Microsoft Hyper-V. Some advanced functions were missing, such as hot-swapping of virtual machines between physical servers and SAN arrays.
“We encountered limitations in management of virtual machines. Things we missed most were functions such as vMotion or Storage vMotion,” says Skowroński.
After deployment we noticed increased flexibility in management of the server environment
For this reason, Ergomedia decided to migrate from Hyper-V to VMware vSphere. For most virtual machines working with Hyper-V, an automatic migration tool was used. Physical servers were also systematically moved to the virtual environment. Some were moved with migration tools, others were reinstalled.
Consolidation of the server environment and replacement of selected servers was the first step. Next was updating VMware vSphere to the most recent version, and virtualisation of all Microsoft applications.
Several new servers were installed at the datacentre to increase the available computing power, but the overall server environment was reduced by 40%. At the same time computing power and available RAM in the vSphere server cluster were increased substantially.
As a result, the IT environment has considerable power reserves and the extensive functionality of the VMware software has supported the development of a complete disaster recovery environment to provide protection against the consequences of failures and disasters.
All servers and disk arrays are grouped in clusters. Virtual machines are backed up on an additional disk array using VMware ESX Server System and a backup agent installed in the ESX Server Service Console.
The new backup method allows for traditional incremental and differential backups. Two levels of redo logs permit maintenance of multiple snapshots and allow for hardware-independent snapshots of virtual machines, yielding true point-in-time copies without the use of SAN features. Also possible is offline backup without having to load the ESX Server that the virtual machines normally run on.
“In some cases we employ a combination of both methods because it’s a way to achieve the best backup results. The major advantage of this approach is that we can back up all the servers in the datacentre. It also allows us to do file-level backup and restoration that gives us more flexibility in the choice of backup software,” says Skowroński.
“After deployment we noticed increased flexibility in management of the server environment, facilitated expansion with additional servers, both physical and virtual, plus a general ease of operation,” he adds.
Now, admins at Ergomedia can spend more time on education and development of competence. Before the migration, their time was entirely used up by administrative tasks for the environment.