2020 needs to be the time for cultural shifts driven by technology; and investment in new network technologies is only a part of the whole answer. While we see disruptive trends in technology, big and small, design-led innovations will drive these changes. This paper highlights the paradigm shifts in design and execution philosophies which will redefine the data centre ecosystem
Data centres are at the core of technological disruption. They have been ever serving the critical role of capacity enabler for enterprises. Right from the mainframe to cloud and captive to colocation, DC ecosystem has evolved massively over the decades.
But is it the end state for data centres or do they still have a long way to go?
We need Super Competent Data Centres for Cloud 2.0.
The following imperatives will pave the way for this to become a reality.
Considering the humongous challenges around scalability and security combined with above-mentioned imperatives, solutions storming for creating competent data centres are absolutely critical
Gartner predicts that by 2025, 80% of enterprises will shut down their legacy data centres. In fact, 10% of organizations already have.So going forward how will your enterprise break the silos of legacy data centres???
One of the key enablers for enterprises to solve these challenges is to implement solutions at the concept level and this could be achieved by ingraining right design elements.
With right planning, enterprises can provision or re-purpose white space and gray space in data centres, implement structured methods for provisioning servers and cables, and prepare their IT environments for the coming wave of virtualization.
Maximising the white and gray space is possibly the most challenging task today. Modern network architectures could use design led innovation to achieve maximum space with optimised costs. Some of the approaches are mentioned below.
The data centre white space is build up using pods. This caters to the IT infrastructure as well as cable management. One of the racks in each row is allocated to networking equipment with a minimum of two fibres to each of the racks in the pod.
Challenging the current raised floor dominance, all networking as well as power cabling could be fed from conduits over the racks instead of under the floor. This, coupled with advanced fibre and cable distribution systems will effectively manage transmissions. Newer concepts like centre racks units capable of supporting extra in-row coolers and connected with under floor water piping can help minimize the number of operational racks, thereby create redundant capacity.
By using these capacity planning processes, enterprises can even measure server and cables utilization trends, analyse future capacity needs, and predict the exact requirement for a given application or service.
IoT sensors installed in data centres can track excessive humidity and control temperature without any human interference. The same applies for connected smoke and heat detectors. They help reduce massive power consumption for management of servers.AI/ML technologies can be utilized to automate and improve operational efficiency, which also includes resiliency and energy usage.
In 2018, Google used deep learning for reducing electricity consumption in data centres’ cooling systems by 40%. Data was gathered from sensors installed on the server racks and then utilized for regulating water pipes for cooling. This could be an excellent use case for many enterprises to improve their data centres efficiency.
Another major improvisation is developing superior cooling techniques. Unlike the traditional approach of locating data centres in cold climates, servers and hardware could be designed in such a manner that can operate efficiently at higher temperatures without impacting reliability
Power backup is another issue, especially for data centres constructed in tier-2 and 3 cities. Frequent power outages lay much stress on diesel generators. One solution could be using rack-mounted UPS system. These can be mounted in racks with width as small as 19" and generally fit in rack spaces. They provide uninterrupted power to the connected devices in the event of a power failure, thereby preventing failure of equipment and loss of data.
Resource-optimized data centres, which could reduce hardware footprint, are need of the hour.
Optimizing servers– Most of the servers are overprovisioned with hardware. To solve this problem, pooled resources can be incorporated allowing servers to share computing resources between systems. These resources can be shared across multiple servers rather than being limited to individual device.
Disaggregated system design - We could also look to inventing a data centre model by enabling a modular, sustainable infrastructure that enables the upgradation of only elements that are lacking in the system rather than re-designing the whole system.
Open source hardware – Enterprises are making hardware used in data centres more efficient, flexible, and scalable. They now opt for the open source hardware projects like the Open Compute Project (OCP), for scale proven hardware designs for their specific usage models. You would never want to buy a traditional server that is designed to run a wide variety of different applications, when all you need is just a hardware that is optimized to perform just one specific task, but can do that task incredibly fast. It uses hardware that is fully optimized to the specific applications enterprises want to run along with optimizing costs.
Virtualization – By inserting a virtualization layer which either uses hosted or hypervisor architecture, multiple OS could be run on one system at the same time. It decreases number of physical servers and increases space utilization efficiency by consolidating servers, You can actually prevent one application impacting the other at the time of upgrades as you have each application within its own virtual server. You can also deploy multiple operating system technologies on a single hardware platform.
Opting for an integrated premise security approach ensures future-proofing the data centre against any type of security breach.
Some of the key elements of this type of a security setup could be:
Optimizing physical security solutions with advanced sensing, deep analytics, IoT, and AIYour focus should be to provide end-to-end solution for achieving critical infrastructure security. It could be a combination of hybrid sensing technologies like fibre, radar, sonar, and lidar with big data and analytics. This would take care of security breaches from a 360 degree angle, be it, sensor-activated fire and natural disaster alarms, motion cameras to detect unauthorised entry or even detecting false alarms.
An integrated CCC to evaluate, communicate, and take action – An integrated Command and Control Centre will enable evaluation of information sent by sensors, communicate it properly to all the stakeholders, and take programmed actions as per the defined situational protocols.
Complementing with cyber security – a data centre security ultimately aims to protect data, which requires installing right set of cyber security solutions on all levels right from installing firewall to monitoring the traffic within the network to mitigating any threat that might have bypassed the firewall.
Apart from above-mentioned security solutions, enterprises can also opt for biometrics, two-factor authentication, and geo-redundant data centres to attain 100% security level.
When you are considering doing away with legacy data centre ecosystem, there are certain step by step approaches you can take to modernize your DC operations:
Install a rack-mounted ATS switch with dual power supplies as modern data centres supply resilience power feeds to maintain uptime and having only one power supply will have negative impact.
Do away with non-standard cabinets as they are not optimal for cooling. Instead use hot aisle containment. Legacy data centres typically utilize manual asset management as well as manual logs for controlling access. This is not efficient for scalability point of view and fully automated digital asset management systems need to be introduced. The access control also needs to be based on a digital system.
SDN allows enterprises to have centralised control over the network and automates service delivery. With the right global connectivity provider, an enterprise can order and provision connectivity services seamlessly across hundreds of data centres, all from a single platform. It also removes the cost and complexity in connecting data centres, while providing a solution for supporting long-term growth.
Data centres are complex ecosystems. There is much to manage starting from power supplies, utility operations, IT infrastructure, cloud platforms, software and DC networks. Currently, all of these are managed in silos. Deploying management and orchestration systems, which allow a single operational picture for all mechanisms is an absolute necessity. Formula is simple – move away from disaggregated to an aggregated approach.
New age data centre management solutions (DCM), specifically designed for a fragmented data centre environment hold the key to provide resiliency, improve uptime, and reduce risks.
Envision a DCM Some of the essential elements of a DCM include:
Data centres need to be imagined and re-imagined as ecosystems and not as establishments. Above mentioned design-led interventions can be extrapolated to numerous other challenges and should be incorporated at the concept stage to create a formidable yet flexible data centre ecosystem to address present day challenges and also to scale them for future headwinds.Download whitepaper