Datacenter Consolidation Strategy

Datacenter Consolidation Strategy 2017-03-23T14:40:47+00:00

Introduction

In the pursuit of delivering IT services that satisfy business demand and drive growth, many IT organizations are now faced with managing an increasingly complex and sprawling IT estate. The evolution of multilayer application architectures and heterogeneous platforms has resulted in a fragmented, silo-managed infrastructure – stretching resources to their limits, and with it much of the IT budget. For many organizations, the only quick and practical solution to meeting demand has been to deploy even greater numbers of servers, storage capacity and network bandwidth within their datacenters, and when the datacenter outgrows its performance and capacity constraints, outsource or build another.

Rising demand

While this approach would seemingly satisfy demand from the business, its long-term efficiency is far from appealing. Ignoring the upfront expenditure required for the new real estate and infrastructure, ongoing power and cooling costs make it an untenable option. Increased global demand has led to soaring energy prices – with recent analyst reports estimating that up to one third of the IT budget is consumed purely by energy costs alone. In parallel, environmental regulations, such as the UK Government’s CRC Energy Efficiency programme, are placing pressure on organizations to reduce carbon emissions. With all these factors in mind, building or outsourcing further datacenter resource without firstly assessing how improvements can be made through consolidation, virtualization or cloud-based options is a wasted opportunity. But although such strategies have been at the forefront of CIOs’ priorities for some years, many organizations will find it difficult to confidently implement a strategy that realizes both its cost savings without compromising on service quality.

Central to any consolidation, virtualization and cloud strategy is the issue of risk and the need to ensure a migration that doesn’t subject the business to any undue disruption. When assessing requirements, a combination of performance, capacity and technology options will need to be taken into consideration, and each has a host of challenges that should be addressed to ensure risk to service is proactively avoided. For example, some initiatives may involve a physical location move that requires cost comparison between sites, network latency and end-user performance satisfaction testing. Other strategies may be motivated by the need to reduce costs, or switching hardware to more energy efficient servers and storage. There are also a number of less obvious issues that need to be addressed too. For example, organizations with users who are currently colocated with the existing datacenter, as these users are effectively ‘hidden’ as far as current WAN bandwidth demands are concerned. But whatever the reason to consolidate, it comes with the inevitable requirement to ensure the initiative is based on reliable facts that de-risk the migration. And it is with this need in mind that advanced analytics provides the robust, quantified facts to ensure all bases are covered.

First and foremost, all consolidation strategies should be based on a robust planning framework which encompasses quantifiable performance and capacity metrics for correctly sizing the new datacenter(s) requirements. However, with many organizations already stretched dealing with everyday services to keep the business functioning, gathering such intelligence is a considerable undertaking and one that requires specialist skills. Traditional approaches of sizing consolidation tend to be based on basic calculations and typically lack accuracy around forecasting userperceived application performance and throughput, post change. Therefore these approaches, due to their relatively high element of risk, require a phased implementation approach, testing each stage of the migration carefully before moving on to the next. Similarly, when choosing to outsource or go to the cloud, careful consideration should be given to any calculations offered by suppliers themselves. While cost-savings from outsourcing and cloud services are tempting, strategies should be based on hard evidence collected in-house from the existing environment, and not grounded purely on conjecture or the merits of the outsourcing/cloud supplier alone.

Taking an analytics approach

Overcoming these issues, advanced analytics enables organizations to maximize the successful outcome of their consolidation strategies by being based on real data. Taking an analytics approach Overcoming these issues, advanced analytics enables organizations to maximize the successful outcome of their consolidation strategies by being based on real data. When assessing requirements for datacenter consolidation and virtualization, a combination of performance, capacity and technology options will need to be taken into consideration, and each has a host of challenges… CUnlike traditional methods of sizing, analytics ensures a robust set of planning requirements by combining real data taken from the existing environment and applying sophisticated scenario modeling to “slice and dice” options under consideration, ensuring the most appropriate decision is selected for implementation.

Baselining the current environment

An advanced analytics approach firstly captures and combines utilization data from the current datacenter estate and end-user/business demand to create a baseline “big picture” model of the current working environment. This baseline model exposes the correlations between performance, capacity, technology and costs to indicate where applied changes can bring the most benefit from both an operational efficiency and cost saving standpoint. By building this baseline model, analytics enables datacenter consolidation strategies to be based on current levels of application performance and capacity, so that any selected initiative can be confidently based on achieving the same if not better standard of service quality, while also achieving its cost and resource saving targets.

Scenario modeling

From building a baseline view of the current position, analytics then enables organizations to understand the before and after picture of options under consideration through the use of “what if?” scenario modeling and change analysis. The benefit of this approach is that organizations can effectively model the likely benefits and impact of each option being considered, providing quantified answers to questions that would be difficult to deduce using traditional decision making methods, such as:

  • What will be the impact of application latency, throughput and user-perceived performance be if we switch datacenter locations?
  • What impact will virtualization make to our running costs?
  • What are the potential cost savings from reducing the number of our applications, and how do we assess which ones are valued by the business?
  • What impact will consolidating our separate datacenters have on our energy costs and carbon-reduction commitments?

Modeling performance

The baseline view establishes the current throughput performance of applications based on current capacity requirements and technology platforms. By capturing utilization metrics at the packet level and applying cluster analysis, analytics identifies common interactions and behaviors that provide a precise understanding on whether applications will be impacted by additional network delay post change, and hence if there is any expected impact on end-user perceived performance. The findings clarified here will ensure performance maintains current levels and that the end-user experience is actively maintained post migration.

Modeling capacity

For determining accurate future capacity requirements, forecasting business growth and disaster recovery requirements, analytics calculates precise utilization measurements from all relevant infrastructure components (network, servers, storage, HVAC, etc.) and their associated running costs. The captured data will cover time of day variations to uncover peaks in demand for each application. By then applying scenario modeling to forecast in anticipated demand from the business (for example, over 3, 6 and 12 months) analytics can then predict each chosen scenario’s outcome, taking into account all influencing criteria such as busy-hour differences, hidden users, technology.

For example, in rightsizing server requirements, analytics models the headroom, technology options, server and rack count required for the proposed change scenarios, accurately predicting the necessary environment. For power and HVAC, analytics can predict server load relationships, energy costs and associated CO2 emission rate of the consolidated environment. Again, the intelligence gathered from modeling capacity from real data provides the assurances that, post implementation, capacity will be able to fulfill forecasted demand.

Modeling technology options

Many consolidation strategies will also encompass some switch to different technology platforms – with virtualization and cloud-based environments being high on many organizations’ wish lists. In terms of assessing requirements and de-risking migration, analytics provides the ability to determine whether a reduction in physical servers will result in the desired cost and management savings, providing robust evidence to support cases for its adoption. Through its powerful modeling capabilities, different scenarios under consideration can be effectively weighed up to identify the most advantageous migration plan – pinpointing the best moves and gains to the organization’s application requirements. For other technology options such as cloud vs. point-to-point, analytics can effectively assess the likely impact on capacity distribution, enabling precise sizing and cost impact to be accurately calculated. Similarly, with comparisons involving MPLS against Ethernet, analytics can determine the impact on application latency and throughput performance between the two, giving precise facts on the most favorable option for the business.

Measuring success and ongoing management

Once consolidation and virtualization efforts are completed, analytics can track their relative progress and business impact by taking regular samples of performance data. The advantage of applying this post consolidation is that it enables organizations to obtain a holistic, end-to-end view of service performance and capacity, establishing a “what normal looks like” model, and enabling trends and developing issues to be proactively identified and addressed. The delivered benefit is that organizations can break away from siloed, reactive infrastructure and application monitoring, and instead, proactively manage the IT estate in terms of business-aligned services, keeping track of costs and eliminating risks from service failure.

Realizing the benefits

While consolidation represents a clear opportunity to redress the balance of stretched datacenter resources, management and running costs, without careful preparation and taking into account the business’ performance and future capacity demands – service quality and expected cost savings can be left in jeopardy. Instead, by using analytics to baseline current performance, capacity and technology options, model scenarios and accurately rightsize the options, IT organizations can be confident in implementing initiatives that achieve their desired objectives and ROI. For the ongoing optimization of services, analytics delivers the quantifiable metrics needed to ensure that ongoing service quality is not only upheld, but that further consolidation opportunities can be readily identified.

aking an analytics approach Overcoming these issues, advanced analytics enables organizations to maximize the successful outcome of their consolidation strategies by being based on real data. When assessing requirements for datacenter consolidation and virtualization, a combination of performance, capacity and technology options will need to be taken into consideration, and each has a host of challenges… Copyright © 2013 Sumerian Europe ® Unlike traditional methods of sizing, analytics ensures a robust set of planning requirements by combining real data taken from the existing environment and applying sophisticated scenario modeling to “slice and dice” options under consideration, ensuring the most appropriate decision is selected for implementation.