The most recent McKinsey&Co study shows that 30% of servers hosted in a datacenter or in data rooms are big „zombie” machines. Meaning, servers that are being kept functional and charge redundantly the monthly bill. Datanet Systems has the necessary competencies and solutions that eliminate “zombie” servers and optimize your operational costs.
Infrastructure orchestration systems in a datacenter, which “harmonizes” management hardware and software solutions is effective not only for removing “zombie” servers but underused servers, optimizing resource allocation. But it is still relatively little used, as shown this year by Gartner analysts, which concluded that 28% of physical servers in data centers are kept plugged in in vain and/or no longer runs any application relevant to the business.
A brief chronological insight on this topic demonstrates that “zombie” machines (also known as “comatose servers”) are being a pressing problem during the recent years:
- In 2015, a study by Stanford University and Athensis Group indicates the same result: one-third of the physical servers in a datacenter work even though they do not serve any business process;
- A third of physical servers in data centers operate while not serving any business process;
- In 2012 Uptime Institute statistics show that one in three servers used in the United States was a “zombie”;
- In 2008 another study McKinsey & Co. shows that 30% of all physical machines are “kept alive” without any purpose.
Multiple loses in datacenter
The conclusion is obvious: “zombies” are hard to kill and charge unnecessarily data centers’ payment bill. Detailed studies show that decommissioning these machines provide significant savings in power consumption. In 2014 experts from the Natural Resources Defense Council were demonstrating with figures and data that if a Data Center would reduce by half the number of “zombie” servers, it could lower your electricity bill by 40%.
Savings are higher but because an active server consumes not only electricity but also space in the data center, and requires maintenance operations and back-up, operating systems licensed, UPS etc. Or, at an average cost of $ 2,000 per year to support operations related to a mid-tier server (according to the Federal Data Center Consolidation Initiative) earnings are consistent.
But beyond cost, the servers kept alive increase the exposure level of organizations against attacks. The vast majority of “zombie” machines receive no software patches and security updates for a relatively long time. This way, they meet all conditions of an optimum backdoor for Advanced Persistent Threat actions and attacks that exploit the “Zero-day” vulnerabilities.
It is not just a “supposition bleak” but a concrete reality – site xDedic out the sale (at an average price at $ 6 per piece) thousands of compromised servers worldwide. In March A.C., xDedic database included 55,000 cars in 173 countries, and two months later had already risen to over 70,000 compromised servers (according to Kaspersky Lab).
To solve this kind of problem, the specialists recommend to datacenter managers to run an audit every three months to identify and get out of function the „zombie” servers. It is a complex action that needs important human resources with advanced competences. This task includes the inventory of every host and identifying its role in the business processes, software, and security updates for every equipment and application.
The process is difficult to be done without the right tools, but on the market exist a vast offer, numerous vendors of IT solutions indicating this problem from various perspectives. For example, the vendors of Data Center Infrastructure Management (DCIM) classic solution sustains that complete inventorying and monitoring of the hardware equipment from data center infrastructure eliminates the „zombie” machines problem.
It is a solid argument, mainly in situations where „zombies” are servers that have been forgotten by the IT administrators. But, on average, „zombies” are servers kept on „idle” for a very long time by the administrators that have not forgotten about them, but avoid to put them out of function and / to use them again. The most frequently invoked reason: the fear of not generating an unplanned downtime or stopping the functionality of an important service of the business. (The saying „Don’t fix it if it isn’t broken” is the best excuse in this case).
DCIM camp begins to lose ground against the application vendors, such as Unified Computing System (UCS), management systems dedicated to data centers designed so as to integrate the hardware and software of IT infrastructures into a coherent assembly, which is managed unitary. For example, Cisco UCS Director does not resume just to the inventory and monitoring operations of the hardware infrastructure, it also provides advanced diagnostic tools in real-time, historical analysis, and usage reports. Cisco solutions provide relief of effort connection and configuration of equipment in accordance with existing infrastructure and running applications requirements and simplify operations provisioning of servers through automation and tools for dynamic allocation of resources, thus leading to an increase in the levels of security and availability.
Cisco has a comprehensive portfolio of management solutions for data centers, which is constantly expanding by the development of new products and functionalities. Such management solution of a next-generation platform is Tetration Analytics, launched this year, which collects through software and hardware sensor, millions of information from the entire infrastructure of a data center, giving complete visibility in real-time. Cisco’s solution is able to detect and learn IT policies applied to the entire infrastructure and behavior patterns of applications, servers, and other hardware infrastructure so it can detect and quickly report any anomaly in their functionality. “Turnkey” delivered platform with preconfigured hardware and preinstalled software applications – not only allows rapid identification of inactive servers but those underutilized, another acute datacenter malady. (According to the Gartner report, in the year 2016, 40% of data centers racks are underprovisioned and 32% of servers are underutilized.) Based on information supplied by Cisco Tetration, data center managers can strategize dynamic optimization with the ability to check them before putting them into practice – through tools integrated into the platform it can be performed “What if” analysis, usage scenarios, historical analysis, etc.
The analysts quoted above warn that technology alone, can not provide a complete response that covers all organizations that are facing the “zombie” machines problem and servers underutilization. For companies that have neither the time nor the necessary resources, to call on a competent partner with expertise in optimization and operation costs in data centers is the best solution.
Datanet Systems, the main Cisco partner in Romania, has a significant number of datacenters architecture implementations, and a complete flexible management solutions portfolio, easy to integrate and operate, that ensures operations efficiency and scale savings in operating Data Centers.