WIXLIB

Solution Brief6 Common Pitfalls toAvoid When MigratingApplications to the CloudBusiness leaders are increasingly relying on cloudtechnologies to earn a competitive edge in today’s globalmarkets. In fact, 81% of companies expect to leverage thecloud to deliver business-critical applications and servicesin the next two years1, spurred by the need to be moreefficient, find better ways to engage and understand theircustomers, and continuously enhance their operations.CIOs and other IT leaders are tasked with delivering theseexpected benefits quickly and reliably, while alsoincreasing their own speed and capacity to deliverprojects of value to the business.But while cloud adoption has hit the mainstream, ITleaders still encounter several challenges when movinglegacy workloads to the cloud. This is partially due to thecomplex nature of today’s hybrid enterprise (Figure 1),where organizations are managing a mix of on-premises,cloud, and SaaS apps, running across a combination ofprivate MPLS links and public Internet transport. A CIO’sability to navigate these architectural complexities iscrucial to ensuring a smooth migration and harnessing thepowerful benefits the cloud offers.1“Simplify and Innovate the way you Consume the Cloud,” Forrester Research, Commissioned byInfosys, October 2014Figure 1 The Hybrid Enterprise

Here are 6 Common Pitfalls that Hinder an Organization’s Ability to SuccessfullyMigrate Applications to the Cloud1. Having an incomplete or outdated view of theinfrastructure4. The proper adoption tracking and performancecriteria haven’t been identified2. Network bandwidth and latency constraints aren’tknown or understood5. Not enough time has been allocated for themigration to take place3. No visibility into how the application is performingthroughout the migration process and beyond6. Application, network, and security teams continueto operate in silosPitfall 1: Having an Incomplete or Outdated View of the InfrastructureThe situationLifting and shifting on-premises applications and gettingthem properly designed, mapped, and then migrated tothe cloud is a complex scenario. The to-be-migratedapplications may consist of hundreds of microservices,make calls to multiple databases, and integrate withcomponents from several different third parties.IT needs to know these components inside and out, andunderstand what services communicate with one another.Otherwise, the application delivery chain will break andsome associated services may experience unexpecteddowntime or performance degradation. Yet many ITorganizations lack a current view of their applicationarchitectures, often the result of a sprawling, dynamicinfrastructure where applications and servicesare continually added or modified, servers that aredecommissioned or consolidated, and other changesand moves that aren’t always clearly documented.Recommended actionsTake principles and lessons learned from data centermigrations and other consolidation efforts, and applythem to cloud projects. Introducing applicationdependency mapping software can significantlystreamline the discovery and planning phase whileproviding IT leaders with a complete and accurate view ofthe entire application landscape, as well as visibility intoperformance issues during and after the migration.Riverbed’s solutionRiverbed SteelCentral NetProfiler is a centralizedreporting and analysis console that combines networkflow data with packet-based performance metrics. Itincludes a discovery wizard that creates applicationdashboards to automate the process of mappingtransactions to their underlying infrastructure so thatapplication definitions and interdependencies areaccurate. NetProfiler can also create service maps,further accelerating the planning process for cloudbased migrations.Relying on a clipboard audit and interviewing applicationstakeholders isn’t enough. To mitigate risk and preventunwanted outages, IT leaders must understand thesecritical, hidden dependencies and validate howapplication components actually talk before theystart the migration process.Riverbed Solution Brief2

Pitfall 2: Network Bandwidth and Latency Constraints Aren’t Known or UnderstoodThe situationMoving applications and data to the cloud imposes newdemands on the corporate network and impacts twoperformance constraints: bandwidth and latency. Notunderstanding the true ramifications on these twoconstraints can easily offset the expected businessvalue of moving to the cloud in the first place.During and after the migration, bandwidth utilizationincreases and traffic patterns begin to vary significantly.Quite often, network links become saturated, which candegrade the end-user experience of both existingapplications and the newly deployed cloud application.Hybrid networking, which combines reliable MPLS andfast Internet links, has emerged as a cost-effective wayto maintain the appropriate amount of bandwidth whileensuring optimal performance for companies leveraginga mix of cloud and on-premises applications. As a result,customers now have options to send a portion of theirtraffic over the public Internet as an alternative toupgrading their MPLS links. Using IPSec technology, theycan backhaul the less critical traffic to the data centerand, therefore, make room for business-criticalapplications over the MPLS links.But in traditional network architectures, where the trafficis backhauled to a central gateway located in thecorporate data center, cloud-based applications end uptraveling a longer distance to reach users when comparedto their on-premises equivalents. Therefore, the time ittakes to complete a business transaction may increase,sometimes dramatically, due to the added latency.This problem can be alleviated by using local Internetbreakouts, enabling direct connections from branchesto the Internet instead of backhauling traffic to a centrallocation. At the expense of additional security costs, theperformance of the cloud application can significantlyimprove as the latency decreases.However, this does not mean that latency completely goesaway. As with traditional on-premises applications, acloud-based application is hosted in a specific location.And in a distributed organization, that location, bydefinition, cannot be close to every user.Recommended actionsStart the cloud migration project with an assessmentphase. Visualize and quantify these end-to-end networkconstraints by testing key business transactions on apre-migration basis. Using predictive analysis and networkmodeling tools can help evaluate current performance anddetermine areas for improvement by pinpointing whereexcessive time and bandwidth are spent within theinfrastructure as the transactions are tested.Once those baselines are completed, explore alternativesto buying more bandwidth when addressing overtaxedlinks. WAN optimization controllers, for instance, canstreamline the movement of large chunks of data,reducing network utilization and associated costs whileaccelerating applications and increasing businesstransaction throughput.Next, consider ways to maximize the usage and flexibilityof today’s hybrid networks. While hybrid networking can becomplex, there are solutions that mitigate that complexityby using dynamic and application-aware path selectionand quality of service (QoS).When appropriate, take cloud-destined traffic direct tothe Internet, eliminating extra hops between the cloudand employees.For latency-sensitive applications, such as MicrosoftOffice 365, or custom applications running in Amazon WebServices or Microsoft Azure, consider WAN optimizationtechnologies that specifically address the needs ofcloud-based applications.Riverbed Solution Brief3

Riverbed’s solutionsRiverbed SteelCentral Transaction Analyzer is theindustry-leading solution for application networkreadiness testing. By running “what-if” scenarios, you cantest critical business transactions to predict the impact ofchange on application response times before you migrateto the cloud and validate proposed solutions—includingWAN optimization—before deployment by assessing theeffects of adjustments made to network and applicationparameters.Riverbed SteelHead is the industry’s #1 optimizationsolution that accelerates delivery of all applications acrossthe hybrid enterprise. With SteelHead SaaS, you canoptimize cloud-based services and SaaS applications,including Microsoft Office 365 and Salesforce.com, byproviding up to 33 times faster performance and reducingbandwidth by up to 97%. And with integrated quality ofservice (QoS) and business-intent path selection,SteelHead provides CIOs the ability to control networkconsumption, as well as what cloud applications usewhich network, by prioritizing business-critical andlatency-sensitive applications over recreational apps.Lastly, the solution secures optimized and non-optimizedtraffic over hybrid networks with standards-basedencryption for added security and regulatory compliance.Pitfall 3: No Visibility into How the Application is Performing Throughout theMigration Process and BeyondThe situationRecommended actionsWhen performance problems do occur—whether asa result of the previously covered pitfalls or anotherunderlying cause—IT leaders need the ability to quicklypinpoint and resolve the issue. However, the cloudintroduces an order-of-magnitude increase incomplexity when trying to mitigate the impact ofperformance problems.Bringing back some level of visibility and control iscritical to maintaining performance and a consistentuser experience in the cloud—throughout the migrationprocess and beyond. Having the proper application andnetwork monitoring tools in place allows IT organizationsto decrease mean time to resolution (MTTR), reducingsupport tickets and cutting technical costs related tomigrating and supporting the cloud application. Andwhen performance issues stem from cloud providers’services, it allows IT to quickly escalate and hold theproviders accountable for agreed-upon service-levelagreements (SLAs).Why? Because now that the application has been migratedor is in the process of being moved to a cloud or SaaSprovider’s environment, IT is relinquishing someadministrative control. The enterprise no longer owns orhas direct access to the infrastructure upon which itsapplication is now hosted. Yet, IT is still responsible for theperformance and security of those services.Making matters more difficult is that IT loses visibility intoapplication performance. The question of whether aproblem stems from the network or the application itselfnow extends to the cloud—perhaps the provider’s servicesare experiencing issues. If it’s on the cloud provider’s end,internal staff members no longer have quick access to thepersonnel who can directly troubleshoot the anomaly.In today’s highly distributed enterprises, where branchesand other remote sites are often the frontline, revenuegenerating portions of the business, having end-to-endvisibility is even more imperative when delivering cloudbased applications and services. These “eyes at thebranches” allow IT to stitch the entire picture togetheracross infrastructure components—WAN optimizers, loadbalancers, firewalls, SDN protocols, etc.—increasing thereliability of critical IT services.Riverbed Solution Brief4