Note: This presentation was designed for 30-45 min time frames, watch your time if you only have 30 mins or drop slides to make the time work.
Standard Oracle safe harbor statement
There are many terms that customers use to describe what they want out of their next data center, these terms define expectations from their future state or the tools and technologies they will use to build this future state. Touch on a few terms that are relevant for your particular event.
This slides sets up the two main sections of the presentation, the text explains the concepts, there are two main paths customers embark on when doing a data center transformation, custom design or one designed from existing building blocks. These two architectures are not exclusive and the same customer may implement both approaches in their environment to solve for different problems.
To help you increase your business agility, Oracle has a complete set of products, integrated to transform your data center and take you to the next stage in your strategy to create an efficient data center.Since you are already familiar with our servers, storage and OS, we will focus on the data center fabric, which connects your servers and storage together. Data center fabric, virtual networking, software defined networks (SDN) are all in the same familt of products and services that allow customer to connect and manage their date center infrastructure more efficiently.
Network fabric is an industry term that describes a network topology in which components pass data to each other through a meshed switching environment where every element can be easily connected to any other elementThe word fabric is used as a metaphor to illustrate the idea that if a data center architect were to draw network components and their relationships on paper, the lines would weave back and forth so densely that the diagram would resemble a woven piece of cloth.
Industry’s first open architecture data center fabric:We support Oracle SPARC servers and x86 systems, including x86 servers from major vendors;hypervisors (VMware, Microsoft HyperV, Red Hat KVM), native OS (Windows, Oracle Solaris, OracleLinux), Oracle VM.Fastest Data Center Fabric:4x faster than competitors (4x server link speed): We use 40Gb bandwidth per server link connection vs. our competitor’s 10Gb.Up to 19x faster VM migrations: We boost application performance and management flexibility by dynamically connecting virtual machines (VMs) and servers to networks, storage devices, and other VMs, allowing you to get up to 19 times faster VM migration.Half the cost of Cisco UCS solutions: (Oracle Virtual Networking + Oracle x86 servers + Virtualization layers) is half the cost of (Cisco UCS + VMware solution)Many customers have seen 50% or more cost savings with OVN. Deployed by hundreds of enterprise customers in mission-critical environment.One-stop Support: Integrated support from applications to infrastructure. So no matter where the problem is in the Oracle red stack, you make one call to Oracle for all your support needs. We can deliver a much higher quality of support and a better overall SLA.
Conventional I/O Expensive, complex… and impossible to change once done.Convergence means:Consolidating infrastructure 70% less complexityAny-to-any connectivity Take one wire and make it do the job of many More efficiency and more flexibility.
Oracle Virtual Networking delivers complete fabric infrastructure for the data center. OVN simplifies complex data center deployments with a wire-once solution and simple software-defined network configuration. Oracle Virtual Networking (OVN) products offer solutions to challenges we discussed earlier:OVN consolidates server connectivity and reduces the number of devices by up to 70%.OVN virtualizes the data center infrastructure with the ‘private virtual interconnect’ in Oracle SDN, allowing connection from any VM or server to other VMs, servers, network and storage devices.OVN enables you to add more services quickly by providing a data center infrastructure that is simple to deploy and easy to reconfigure for your cloud deployments.You get predictable performance for mission-critical applications with our granular QoS controls for network and storage access.We also provide security in multi-tenant Cloud environment, so you can protect your customers’ sensitive data.
The Oracle Virtual Networking product family consists of 4 major elements. One is the hardware that you see on the top left of the picture here, which are the Oracle Fabric Interconnect boxes. You can select between the 4U-high system with 15 I/O module slots and the 2U-high system with 4 I/O module slots.There is a management software element, which is called Oracle Fabric Manager, shown on top right of the slide. It consolidates I/O management across thousands of servers to a single console, providing complete control of your data center fabric. You log in and you can basically manage the entire infrastructure from a single pane of glass.Going clockwise, bottom right of the slide shows the performance monitoring product called Oracle Fabric Monitor, which allows you to get a consolidated reporting view of data traffic usage across every virtual NIC, virtual HBA, every port, and every chassis that you have in your infrastructure. Finally, the Oracle SDN software allows you to take servers and bind them to each other using private L2 networks and getting extremely fast application performance, such as VM migration.
Oracle Virtual Networking provides a converged solution, which means that you can put Ethernet and fibre channel on the same wire. But that does not mean that you compromise. Because the fabric is lossless and highly reliable, those fibre channel frames are guaranteed to deliver from the server into the fibre channel fabric in the same manner as if you have a native HBA device on the host. You get full line rate on any packet flows.Oracle Virtual Networking has been proven to be an enterprise solution. There are over 300 customers worldwide using this product family.
(Notes: this slide has builds. There are detailed speaker notes to explain the slide. Please continue to read the speaker notes on the next page, which is marked as ‘Hide’ in Slideshow mode).If you look at the world today, this is the picture, right? Every server has its own network adapters, its own storage adapters. This architecture is inflexible and not scalable. It does not allow you to make changes to the architecture easily. With Oracle Virtual Networking, we can get rid of those network devices and storage devices you have on the server. Instead every server is connected with a one-time point-to-point, high-speed, high bandwidth, extremely low latency link into these two hardware boxes you see in the middle of the picture, which are called the Oracle Fabric Interconnects. And the reason we show two boxes, instead of one, is that this architecture gives you high availability.Now what you do is install some drivers on every server, and the two Oracle Fabric Interconnects will contain all your network adapters and storage adapters. In this picture, you see that each one of those Fabric Interconnects has four I/O module devices in the bottom. So those are the devices that connect into your 1 Gig or 10 Gig Ethernet fabrics or into your 8 Gig fibre channel fabric. And what we do is we take a single 8 Gig fibre channel port for example, bind into a switch, and we virtualize that port. We make that port look like 64 fibre channel HBA devices. Similarly, if we take a 10 Gig Ethernet link, bind it into a switch port and we make that one 10 Gig link look like 128 Ethernet NIC devices. So essentially these two fabric interconnect boxes that you see in the middle contain all your network adapters and storage adapters for a rack or two of servers. Once the systems are installed and set up, you never have to go in and touch the server again. Using our management software, you can simply go and add as many network ports or storage ports you need on any one of the servers that are bound into this fabric. For example, you are running Oracle virtual machine on that server in the left of the picture and you need to install 12 Ethernet devices and 4 fibre channel devices. You simply go into the Oracle Fabric Manager console and you say, give this server 12 Ethernet NICs and 4 fabric channel HBAs. They show up on the host and the beauty of the solution is that as far as the server is concerned, it is as if somebody has gone in and plugged in 12 Ethernet adapters and 4 fibre channel adapters. So, they look and function and behave just like native adapters.
Situation:Bluelock is an enterprise cloud hosting provider and is one of only a handful of companies certified by VMware to offer VMware vCloud Datacenter Services, a service that VMware calls an “IT revolution for millions of existing datacenter applications.” With Bluelock, clients can fluidly migrate their VMware-based applications to Virtual Datacenters hosted in the public cloud. Bluelock Virtual Datacenters help companies get started with projects quickly, while delivering the freedom to change their minds as IT needs evolve.Challenges:The first was to reduce infrastructure complexity. Bluelock’s director of IT commented: “To our clients, our service is all about flexible capacity. It’s essential that we respond quickly to their changing needs, so our infrastructure must be adaptable and scalable.” Traditional server I/O did not meet this goal. Every server needed 10 physical I/O connections, and for every 500 virtual machines deployed there were 50 network devices. This resulted in complexity that reduced flexibility and increased cost. Furthermore, because it was hard to find and fix trouble spots, it became increasingly difficult to meet their aggressive uptime objectives.The second challenge was to decrease the time required to migrate a virtual machine from one host to another. With traditional I/O it took 60 to 120 seconds per VM, which meant that it could take hours just to evacuate all of the VMs from a host and to then move them back. In day-to-day operations, this duration limited the number of virtual machines that could reside on a single server, which in turn limited the efficiency of the overall environment.The third challenge was change management.Traditional I/O not only added cost and complexity but also made it very difficult to add I/O capability or new resources. Any modifications that required infrastructure changes were time consuming, thus increasing the total cost of adding new capacity or services.
Solutions:Using Oracle Virtual Networking, Bluelock got a significantly simplified infrastructure, requiring just 2 I/O cables per server rather than 8. The OVN solution also reduced the number of I/O cards per server from 4 to 1, and slashed the number of network devices needed. “The number of virtual machines we can host per network device has now increased from 10 per device to 400 per device,” per the IT Director.a vMotion event now takes just 20 seconds per virtual machine, rather than 60 to 120 seconds, due to the high-throughput server-to-server communications that OVN enables. Fast migration allows Bluelock to evacuate servers far more quickly, which allows the company to run more VMs per server and still meet customer SLAs. Because new servers can be added with just two cables, capacity expansion is accelerated, reducing the need to add capital equipment far in advance of demand. And because configuration changes can be accomplished entirely in software, resources can now be re-assigned in minutes. That means Bluelock can quickly accommodate the needs of their largest customers. Configuration agility has also allowed Bluelock to streamline services such as outsourced disaster recovery. With OVN, it is possibleto quickly replicate a customer’s own environment for rapid failover. Results:40X more usable bandwidth per server 17X more virtual machines per hardware device8X greater virtual machine capacity, etc…
Section divider slide, sets up the second section of the presentation using engineered systems with a focus on OVCA to deliver dynamic data center services with converged infrastructure systems
The days of rack and stack computing are rapidly coming to an end, the traditional method of the customer as the prime integrator, lead R&D consultant and ongoing change manager are decreasing in their value to the business. The business wants IT to be great at delivering timely services and providing a flexible environment for the business to explore new sources of revenue or better ways to serve their customers. The only large segment of IT spending that is growing is converged infrastructure, customers are using converged infrastructure (also known as integrates systems or integrated infrastructure) to shift the R&D and integration spend from their books back to their vendors. CI provides many great customer advantages in creating a building block architecture for a dynamic data center.
Oracle have been delivering Ci systems before the term converged infrastructure was adopted by the market, we started off with the premise that we could build and integrated HW & SW system that could run the oracle data base better, faster and cheaper than any other system on the market. This became exadata, our longest running and most successful engineered system. In the market definition of CI, you need to have, servers, storage, networking and management in a single package. All of the systems in our ES portfolio fit into the CI market. Until this year all of out engineered systems were purpose build for a specific application or set of workloads (focus on a couple relevant to your event) in August of 2013 we announced a new type of engineered system, one that was targeted at a broad set of workloads across the data center core and that could be use as a true building block for a dynamic data center implementation. This new system is called the Oracle Virtual Compute Appliance and it’s complimentary to the other engineered systems and can run all of the workloads that don’t fit into a specific purpose exa or supercluster system.
The core value of OVCA is ease of use and accelerating the deployment of customers services. The system is very simple to use, just turn on one of the management servers, it boots up the rest of the rack, checks firmware, installs virtualization software, provisions storage and networking and delivers a platform to customers that they can begin provisioning their applications in under an hour. In our launch video our SVP of development had the complete infrastructure up and configured with a oracle 12c data base deployed in about 52 minutes. This is accomplished with the built in software defined network, advanced orchestration software and oracle templates
The virtual compute appliance is able to run all the common workloads you have in your existing data center and is a great platform to base your new dynamic data center designs around. It can run most versions of linux and includes licenses for Oracle Linux, all recent versions of windows and is microsoft virtualization certified as well as Solaris applications that run on an x86 platform. This slide is a build
Overview slide of the inside of OVCA, don’t spend much time on the internal components, the value is in the integration. That said all of the components have been deployed in large scale cloud environments so all the elements that make up OVCA have already been battle tested and perform to the most stringent SLAs our customers have.
One of the unique features in OVCA is the build in SDN or software defined network, this delivers a system that never needs physical changes to add new networks, reconfigure networks or connect networks, this covers all IP networking and storage networking for an extremely flexible system that allows for many different types of application workloads to be deployed on the same box and for customer to move work load around their systems as their needs change. The system also includes hardware for future expandability that will be delivered through software updates. The system is fully pre cabled so even if you don’t start with a full rack configuration expansion is as simple as sliding in a new compute node and attaching a few cables that are already in the rack.
Customer that want to deploy a larger data center environment using OVCA and other engineered and best of breed oracle systems can use Oracle Enterprise manager to orchestrate the management and provisioning of all of their system. EM provides all of the virtual machine and application provisioning as well as cloud functions for self service and metering and chargeback. This optional but very powerful tool is essential for anyone leveraging multiple types of oracle infrastructure in their data center.
The Oracle templates and virtual assembly builder are key technologies in OVCA that allow you to deploy Oracle applications and applications efficiently and repeatably. Oracle creates templates for over 100 different Oracle software products so customer can save the basic setup and deployment time event time they need to create a new instance of the SW. The templates and assemblies are also open and extensible, for example if you deploy a custom application every time you deploy RAC, just start with our RAC template and add your SW to it to create your own master template.
At Oracle, we always knew that Oracle VM Templates can save significantly on deployment time.To validate this, we commissioned the Evaluator Group to conduct a test in their independent labs using only publicly available Oracle documentations.In this lab validation, the Evaluator Group deployed a 2 node Oracle Real Cluster using Oracle VM templates.The group also deployed the 2 node cluster Oracle RAC on VMware vSphere.The resulting time savings was 10x with Oracle VM taking about 3 and a quarter hours vs. with VMware vSphere taking over 39 hours.
Slide covers most of the points in the bullets, the key take aways here is that expansion is simple, need more compute add more nodes, need more storage, we can connect to any existing or new storage to increase the data storage needed to support your OVCA deployment
OVCA systems come with a ZFS Storage Appliance to manage their internal data, boot images, etc., but what about the large amounts of data that you need to access to run applications on your virtual machines?Oracle’s ZFS Storage Appliance is ideal for that too. It is co-engineered with OVCA to maximize performance and efficiency while at the same time reducing deployment risk and TCO. As an Application Engineered Storage expansion to OVA, ZFS offers these capabilities:Extremely high throughput for any type of applications and workloads you are running on your VMs. It has optimizations for IOPS-intensive workloads like OLTP databases, and for bandwidth-driven workloads like data warehousing, BI Analytics, and video processing. And it’s powerful enough to concurrently run a diverse set of workloads.You will also have to work hard to run out of capacity with the ZFS Storage Appliance. It comes in multiple sizes with raw capacities for application and database data that expand from 20TB to over 2.6PBThe ZFS Storage Appliance also comes with the world’s best storage analytics, which allow you to visualize and drill down into specific workloads to understand where bottlenecks occur and why. It can even allow you to examine and manage the storage aspects of OVCA environments all the way down to the VM level.And finally, the ZFS Storage Appliance has the core Application Engineered Storage features since it is co-developed together with other Oracle hardware and software to deliver specific capabilities not available in third-party storage systems. For instance, it reduces risk by automating storage management from Oracle Database and having fewer storage systems to integrate and manage, by providing leadership fault-monitoring and self-healing capabilities, and by simplifying setup and management with DTrace Storage Analytics. It reduces complexity because its large memory and cache- based architecture is more efficient in serving the I/O from large virtualized environments, and unique features like Hybrid Columnar Compression reduce the amount of storage needed for data warehouses built using Oracle Database. And, it enables you to lower TCO because you need fewer systems that cost less and are easier to manage.When you add all of this up, the ZFS Storage Appliance with its intelligent caching gives you the IO performance of Flash but with the cost and scalability of Disk-based storage solutions.
Wrap 1 of the presentation, recap what’s great about OVCA and the rest of the engineered system family of products
Wrap 2 of the presentation, Call to action, get started now, there are newer technologies and systems that can help form the base of your dynamic data center, don’t just buy what you always have, investigate your options and question what you really need to build vs buy as a solution.
"Deploying a Dynamic Data Center" - Karol Płuciennik, Principal Sales Consultant, Oracle Polska