Authored by Dan Lahl & Amit Sinha
I am pleasantly surprised that after years of dismissal of the cloud in its many different variations such as virtualization, Software as a Service, Data as a Service, Public Cloud, Private Cloud and Hybrid Cloud, – Oracle – is now finally waking up to the fact that today’s enterprises DEMAND cloud capabilities from its data management vendor. We at SAP, along with the many customers who have been shouting for cloud capabilities from Oracle, eagerly await Oracle 12c, announced at Oracle Open World this week – to be delivered sometime in 2013.
And we at SAP would also like to welcome Oracle customers who want to take advantage of proven and mature multi-tenant and cloud capabilities TODAY (not promised in the future). For over 25 years, SAP customers have benefited from the ability to host, manage and isolate multiple databases in a multi-tenant cloud environment tor transactional, analytical, mobile and realtime data management applications. SAP has enabled our customers to offer separate databases, securely isolated from other user’s data, but simple to manage and back up. These separate databases are not only multi-tenant, but can also be virtualized and tuned to balance the load across servers in a public, private or hybrid cloud environment.
From the Virtual Server Architecture of SAP Sybase ASE, to Virtual Datamarts in SAP Sybase IQ to the multi-tenant capabilities of SAP Sybase SQL Anywhere on demand edition, SAP’s real-time data platform is delivering for the cloud today. This, added to SAP HANA’s REAL in-memory capabilities vs. Oracle’s FALSE memory claims for Exadata X3 (see Vishal Sikka’s blog here: http://ow.ly/eaEoj ), makes the SAP real-time data platform the right choice for customers integrating applications and data management going forward. Read on to see how SAP Sybase data management portfolio – has been for years the world’s first multi-tenant and cloud database provider.
SAP Sybase SQL Anywhere ON Demand
SAP Sybase SQL Anywhere, On-Demand Edition is a multi-tenant, cloud-enabled database designed primarily for use in software-as-a-service (SaaS) environments. On Demand Edition’s main distinguishing feature is its ability to host, manage, and isolate multiple databases across a database cloud comprised of many individual host machines. These host machines together form a database cloud that is capable of running multiple, individual tenant databases that can be plugged into individual servers in the database cloud. Tooling is provided to allow an operator to perform bulk actions on multiple tenant databases in a single action. All tenant databases can be backed up in a single operation, however all databases may still be restored individually. The tooling also allows an operator to manage the underlying SAP Sybase SQL Anywhere, on-demand edition software, and to roll out updates to logical groupings of host machines.
Each tenant database in the system is completely isolated (file-level, meta-data, credentials, encryption) from all other tenant databases. Tenant databases can be further process-level, memory-level, and machine-level isolated if required (determined on a tenant-by-tenant basis). Furthermore, individual parts of a single tenant database may be isolated into separate files. This isolated, multi-tenant architecture has been the basis of SQL Anywhere for over 20 years. This isolation provides the security required to implement multi-tenancy at the database level, and mitigates the risks of implementing it in the application. Furthermore, the tenant isolation allows tenant databases to be easily moved between servers in the cloud for the purposes of load balancing.
An SAP Sybase SQL Anywhere cloud may be elastically scaled during runtime by adding and removing host machines. The cloud is a shared nothing architecture allowing the host machines to be geographically distributed (including between multiple public IaaS providers). A cloud may be comprised of commodity servers (with as little as 1GB of RAM), as well as large servers. The cloud is licensed based on the number of cores, and is available in both a perpetual model, and a per-core-hour utility model. The SAP Sybase SQL Anywhere engineering team has successfully setup and run a single cloud powering thousands of individual tenant databases spread across hundreds of host machines. Moreover, SAP SQL Anywhere on demand edition is available today!
SAP Sybase Adaptive Server Enterprise
SAP Sybase Adaptive Server Enterprise (ASE) has supported the concept of logical databases since its inception in 1987. For 25 years ASE servers have allowed end users to share a single database server instance with up to 65,000 logical databases which have independent security, transaction log, maintenance, and backup. This makes it easy to consolidate many users and independent databases on the same system, dynamically sharing the load among its processors. Yes, you can have up to 65,000 ASE servers in an ASE “container”. How is that for multi-tenant!
ASE’s Virtual Server Architecture is designed for cloud and highly consolidated environments where resources (CPU, memory) are optimally shared in these highly virtualized environments. ASE has always supported multi-tenancy as an intrinsic part of its design. It also supports application specific schemas (ASE databases), and provides isolation between application as well as users.
With ASE, managers have the ability to manage SLAs between different application by leveraging ASE’s capabilities to prioritize and manage server/machine resources for specific applications/tenants. Schemas (databases) can be independently administered, recovered, backed up, restored etc. Security capabilities support multi-tenant ASE servers. Strong privacy control capabilities, designed to meet the strict demands of military and the financial service industry are implemented. Encryption technology can be added to further enforce separation of data access a columnar level.
SAP Sybase IQ
SAP Sybase IQ supports Virtual Data Marts in analytical environments, providing cloud capabilities since 2011 that many data aggregators from finance to telecom to retail provide as a cloud service to their customers today. With Sybase IQ PlexQ, IT departments can implement virtual data marts in order to optimize the experience for different applications and user communities. Sybase IQ PlexQ dynamically balances query workloads across nodes on the grid for massively parallel processing of complex analytics at speeds up to 100 times faster than traditional data warehouses. Unlike current shared-nothing MPP architectures that are optimized for certain workloads, the combination of Sybase IQ’s columnar data store, shared-everything grid, and logical servers ensures top performance for any query, any workload over any dataset.
Many enterprises suffer from data mart sprawl as individual departments seek dedicated support for specific analytics and reporting purposes. A staggering 35%-70% of data mart costs are unnecessary according to research by Gartner. Consider that 20-45% of any data mart is stored elsewhere, and utilization of data mart server machines is often well below 10%. This results in overly complex data architectures that incur extra costs for duplicate data, unnecessary effort to replicate and load data, underutilized hardware, and multiple versions of the truth.
With its unique PlexQ architecture, Sybase IQ offers a different approach that bands together physical servers into virtual groups that can be dedicated to certain users or processes. Logical Servers within a PlexQ grid allow IT staff to create many virtual data marts on one physical grid. Each Logical Server can have its own set of compute resources, separate data if so desired, and separate security. The performance of each virtual data mart is not impacted by any others on the grid. This isolates the impact of different analytic workloads, ad hoc queries, and users from each other making it easy to implement information-as-a-service underlying either private or public clouds
Virtual data marts have all of the benefits of a physical data mart while reducing overall data mart costs by up to 70% by eliminating duplicate data, simplifying maintenance, and reducing hardware and software costs. Furthermore, each virtual data mart can access any of the data on the grid, if given permission, and additional compute resources can be dynamically reassigned as workloads change.
In summary, what we see from the announcement at Oracle OpenWorld 2012 with Oracle 12c is an attempt to put old wine in a new bottle. The concept of Virtual databases in a container database is nothing but managed tablespaces with its own schema that has existed in SAP Sybase products for a long time. What do you think? How should SAP respond? Thoughts are welcome.