Citrix cloud platform - Journey to IT-as-a-Service

Editor's Notes

1. Accelerate cloud results with the only solution proven to orchestrate both existing and cloud-native workloads.
2. Mega trends impact IT and become their challenge
3. Intro: In fact, when organizational leaders are asked why they are making decisions about private Cloud… Key Points: InformationWeek surveyed their readers and received responses from 242 at organizations with 50 or more employees and asked about the reasons for considering a private cloud. These are among the top reasons. The top 2 reasons are cost-related. 61% are looking for significant operational cost savings 54% are looking for significant capital cost savings 48% feel that they would consider a private cloud if it gave them a compelling technical advantage 25% feel that having a private cloud would give them a lower cost of entry into other markets 16% look for product integrations and management tools when considering private clouds Illustration/Anecdotes/Proof: Source: InformationWeek 2014 Private Cloud Survey Transition: While the benefits of private cloud can translate to cost savings, they are looking for significant direct savings when considering these types of projects, they are also applying non-technical, business-level metrics such as ability to enter new markets and interoperability when considering private clouds.
4. What is CloudPlatform? Citrix CloudPlatform, powered by Apache CloudStack, is the industry’s only cloud orchestration platform proven to future proof your cloud by running today’s traditional workload and tomorrow’s cloud era workload. This mature, turnkey solution is based on an open and flexible architecture that is trusted to power the world’s leading public and private clouds. For example It scales to over 40,000 servers 35,000+ Open Source community members 1000+ Citrix Ready Cloud ISV partners 150+ production clouds
5. CloudPlatform is designed to work within your environment, leveraging your existing infrastructure or allowing you to choose best of breed components for compute, storage and networking. CloudPlatform supports all major hypervisors, a large number of storage options and includes comprehensive networking, plus includes the ability to directly interact with a number of virtual and physical networking components.Most cloud buildouts are not greenfield datacenters. The majority of enterprises and SPs are looking to turn existing virtualized infrastructures into clouds. The flexibility of CloudPlatform is key differentiator in that it works with existing infrastructure (physical and virtual) without forcing a platform overhaul.
6. The easiest way to think of CloudPlatform is to compare it to server virtualization. Where XenServer creates an abstraction of server resources (CPU, Memory, etc.) to increase density and control of a server, CloudPlatform creates an abstraction of datacenter resources (Compute, Network, Storage) to increase scalability and control of a datacenter. This is a quick illustration of the CP architecture outlining the core components in blue. As you can see, it is a comprehensive set of technologies that overlays existing components and interacts with existing systems to act as the cloud operating system for your datacenter. We’ll dive into some of the core features in a minute…. Developer API CloudPlatform provides an API that gives programmatic access to all the management features available in the UI. This API enables the creation of command line tools and new user interfaces to suit particular needs. CloudPlatform can translate Amazon Web Services (AWS) API calls to native CloudPlatform API calls so that users can continue using existing AWS-compatible tools. This translation service runs as a separate web application in the same tomcat server as the management server of CloudPlatform, listening on the same port.
7. HA VM HA to protect against physical server failures VM placements with anti-affinity for higher resiliency Non disruptive Operations Live migrations for routine maintenance or server upgrades Dynamic Scaling of resources as workload resoruce needs grow
8. comprehensive networking, plus includes the ability to directly interact with a number of virtual and physical networking components. By using best-of-breed networking components from different vendors, the cloud can be designed for performance and scalability, delivering SLAs for demanding production workloads.
9. This picture illustrates how the different XD components are deployed on a CloudPlatform based private cloud. To design for high availability of desktops, a XenDesktop Site should be extended across two AZs in geographically split data-centers. The split data centers should typically be few miles apart to ensure very low latency with independent power lines and internet connectivity. In the event that physically separate data centers are not available, a single data center can also be logically split into two to support the above architecture. One AZ acts as a backup for XD control components with replication of DB from the active AZ. The multi-AZ design also allows for scalability to address future growth. Desktop Deliver Controller (DDC) A pair of DDCs and Database are deployed in the primary CloudPlatform AZs Another Database is deployed in the secondary AZ. This DB is actively replicated from the database in the primary AZ Another pair of DDCs are also deployed in the secondary AZ, but continue to use the DB in the primary AZ until it’s no longer available and the backup DB is promoted to be the primary The DDCs and DB should be ensured that they are deployed on different physical hosts using CloutPlatform anti-affinity feature    AD Domain Controller (AD DC) There is at least one AD domain controller in each of the AZs   StoreFront (SF) There is at least one StoreFront in each of the AZs   NetScaler (NS) There is at least one NetScaler in each of the AZs that servers as a access gateway as well as a Load Balancer GSLB is configured across the Netscalers in the AZs to ensure high availability of StoreFront (and thus XD service)
10. Fundamentally there are two different types of workloads or applications that are run in today’s cloud environments—traditional and cloud-native. Each of these workloads have uniquely different architectural characteristics and distinct set of requirements around infrastructure and availability. Traditional workloads The majority of today’s existing enterprise applications that live in the datacenter fall into this category. They include, for example, such SAP®ERP, Oracle® database apps and Microsoft® Exchange. These are typically client-server apps built to run on a single server or a cluster of servers and databases. Traditional applications achieve scale by scaling up, that is, by increasing the size of the application and database infrastructure. these applications typically serve tens of thousands of users and hundreds of concurrent sessions. Enterprise workloads are also traditionally designed to run on reliable, enterprise-grade hardware; And reliability is ensured through complex enterprise technologies such as network link aggregation, storage multi-pathing, virtual machine (VM) high availability. Sophisticated backup and disaster recovery procedures are put in place to handle the unlikely scenario of hardware failure. Choice of infrastructure components include: Commercially supported hypervisors such as Citrix® XenServerTM or VMware® vSphereTM High-performance SAN devices for VM image storage Traditional physical network routers, firewalls and layer 2 switches VLANs to isolate traffic among servers and tenants VPN tunneling for secure remote and site-to-site access through existing network edge devices   Cloud-native workloads Cloud-native workloads include the new generation of apps associated with cloud computing, for example gaming and mobile apps, High Performance Computing (HPC), Big Data, social apps and batch processing. They achieve “web scale” by “scaling-out” across many commodity grade nodes of compute, networking and storage, to cost-effectively support the load generated by many millions of users. Built for hardware that isn’t expected to be resilient, these workloads are designed with the intelligence to handle the failure of any given node. Disaster recovery is achieved via a multi-site failover design and geographic load balancing. Infrastructure components typically include Commodity and open-source components : An affordable hypervisor such as XenServer, Xen or KVM NFS volumes and inexpensive local disk for VM; And Object store for images. Software Defined Networking (SDN) and L3 Security Groups used for workload and tenant isolation Elastic Load Balancing (ELB) or Global Server Load Balancing (GSLB) is used to redirect traffic to servers in multiple availability zones Take away here is that CloudPlatform can deliver both these workload types.
11. Traditional or Legacy Workloads – SAP/Oracle/Sharepoint – have different scale and fault tolerance needs than Cloud-Native workloads like those hosted by Amazon, Facebook, Google and Zynga. Legacy workloads are generally supported by an infrastructure built on proprietary physical gear leveraging SANs for storage and physical networking gear for firewalls etc.
12. In contrast – most cloud-native workloads are often supported by commodity hardware and software and leverage block and object storage, SDNs & security groups for networking and isolation. They assume failure and leverage multiple zones or locations for failover, not additional physical devices.
13. Citrix CloudPlatform is the only IaaS CMP (Cloud Management Platform) that is designed to support both types of products from the same management plane. This allows your environment to embrace one workload today and grow into the other as needed. This allows the reuse of your existing assets and applications while accounting for growth over time from within the same management product.
14. Here is a quick look at a few of the customers who are running Citrix cloud offerings today in their environment, BT and TaTa on the public cloud front, Spotify and Edmunds.com are some of our web 2.0s and we’ve seen a lot of growth in the enterprise and education market over the last year with the likes of Nokia and Autodesk.
15. About the Customer Top 10 engineering and project management company Challenge/Pain Develop a private cloud that would increase efficiency and provisioning while meeting complex requirements and strict security and compliance rules Key Benefits Improved project support by moving provisioning process closer towards the customer Increased efficiency and planning by delivering the service on-demand vs. over provisioning Provides governance/consistency layer for apps Improves quality of service by eliminating variance in applications and infrastructure
16. Challenge Multiple datacenters, departments, offices and countries to support Manual IT service and provisioning was slow and inefficient, Usage metering by department Multi-hypervisor requirements Goals Standardize on app and services delivery Cloud needed to fit into existing infrastructure and investments Solution CloudPlatform for multi-hypervisor IaaS CloudPortal for self-service, metering, business unit isolation End goal of enabling a comprehensive ITaaS model What key business drivers/initiative(s) did the customer want to address that lead them to the solution?   Needed to consolidate, standardize, and orchestrate multiple infrastructure technologies across multiple/four data centers. Needed to standardize application delivery. More than 1200 applications, receiving requests for specific services from IT users representing 20 separate divisions. The list of applications covers multiple categories including XenApp and XenDesktop. The majority of applications run on Windows OS’s, in addition to Oracle DB’s running on bare metal -- all of which are available in all four data centers. Today these IT users (~ IT division heads) request services and applications, and then manually provision these systems for their end users (employees), logging in and creating applications in a specific data center based on the location of the division. The applications have to be launched in the context of the division to respect isolation, metering, etc. The application instances today don’t have any standardization when it comes to CPU/RAM. The initial launch of the cloud services will have standard compute offerings (2CPU, 2 GB to 8CPU 32 GB and all configurations between).   Challenge Multiple datacenters, departments, offices and countries to support Manual IT service and provisioning was slow and inefficient, Usage metering by department Multi-hypervisor requirements Goals Standardize on app and services delivery Cloud needed to fit into existing infrastructure and investments Solution CloudPlatform for multi-hypervisor IaaS CloudPortal for self-service, metering, business unit isolation End goal of enabling a comprehensive ITaaS model What key business drivers/initiative(s) did the customer want to address that lead them to the solution?   Needed to consolidate, standardize, and orchestrate multiple infrastructure technologies across multiple/four data centers. Needed to standardize application delivery. More than 1200 applications, receiving requests for specific services from IT users representing 20 separate divisions. The list of applications covers multiple categories including XenApp and XenDesktop. The majority of applications run on Windows OS’s, in addition to Oracle DB’s running on bare metal -- all of which are available in all four data centers. Today these IT users (~ IT division heads) request services and applications, and then manually provision these systems for their end users (employees), logging in and creating applications in a specific data center based on the location of the division. The applications have to be launched in the context of the division to respect isolation, metering, etc. The application instances today don’t have any standardization when it comes to CPU/RAM. The initial launch of the cloud services will have standard compute offerings (2CPU, 2 GB to 8CPU 32 GB and all configurations between).  
17. About the Customer Online Gaming company, focused on deploying production applications across multiple clouds using a Hybrid cloud environment to reduce costs. Challenge/Pain Build hybrid cloud to support both transient workloads as well long running workloads. Solution was based on using CloudPlatform for Private Cloud and leveraging RightScale, one of Citrix partners, for Hybrid cloud deployment. All of the new workloads were placed on Public Cloud using RightScale. Once the infrastructure requirements for the workload are identified, this customer would bring back the workload into a Private Cloud environment. Key Benefits Workload deployment across multiple Clouds Cost savings by having the ability to easily move applications
18. Korea Telecom   Customer Quote “Cloudstack’s flexible yet performance-enhancing architecture allows ucloud to deliver stability, security and convenience that make it more agile and usable at a price that is 40 percent lower than Amazon Web Services.” - JS Suh, senior vice president of KT   About the Customer Largest landline operator and second largest mobile operator in Korea   Challenge/Pain 1) Choosing the appropriate cloud technology vendor was critical to the success of KT’s first public cloud offering, ucloud. In today’s economy where enterprises are being forced to do more with less, infrastructure cost has been becoming an increasingly large problem. Cloud computing services have huge benefits for IT services, but the cost associated with building, deploying and managing a public cloud can be daunting. It was imperative that KT ucloud provide clients with high performance technology at a fraction of the cost.   2) Acquiring a solution that can integrate with existing infrastructure technologies.   Why Citrix CloudPlatform? Seamless integration with existing infrastructure provided KT with reliable cloud architecture to build new services and offerings Ability to monitor and meter usage volumes allow users real-time visibility into their cloud resources Provides auto-provisioning, auto-scaling allowing users to run their system instantly and delivers on-demand services as needed Dramatically decreased time-to-market   Key Benefits Achieved performance-proven, cost effective platform (cost: 40% lower than Amazon Web Services) Decreased time-to-market with a production-ready system that differentiates itself by tailoring solution to their customer’s needs Minimized overhead integration costs and complex deployment schedules