Sybase Replication Server Update Focuses on Real Time
With the rise of right-time BI and the advent of a culture of what might be called mass (data) consumption, replication is more important than ever, as a new update from Sybase attests.
- By Stephen Swoyer
- May 27, 2009
It's been nearly two decades since its release, and Sybase Replication Server is still going strong -- and it's evolving even today. Moreover, with the rise of right-time BI and the advent of a culture of what might be called mass (data) consumption, replication has arguably become an essential technology.
It's appropriate, then, that Sybase's latest Replication Server update (to version 15.2) announced last week supports a DSI Bulk Load option (which first shipped in a service update for Replication Server 15.1), support for non-blocking commits, SQL statement replication, and an improved manageability feature set. The aim, says Bill Zhang, product manager for Sybase Replication Server, is to augment Replication Server's performance and feature set that addresses several new business cases.
The first of these, he indicates, is a push for real-time reporting.
"We have customers who want to use Replication Server to create an exact replica of their production database, then run their reporting off of that replica database," he comments. "There's also increasing demand to enable something called a 'real-time data warehouse.' If they're a Sybase IQ customer, they'll say, can you help me load data from a Sybase database into a data warehouse in real-time? So that's also important."
The push for real-time reporting is of a piece of what Zhang describes as "a couple of mega-trends" that have recently come to the fore. "We're seeing that data volumes have increased dramatically compared to 18 years ago. Because [companies are] serving worldwide clients, their operating windows [for replication] are shrinking. They have more data to deal with and less time to deal with any operation. The system software has to improve performance in order to meet those requirements," he explains.
Secondly, says Zhang, shops are deploying applications on a global basis, such that a financial application in Tokyo must be able to "talk" with a similar application in London. "Replication is a very mature technology and [customers] see it as ideal for [use cases like] this," he indicates.
The not-so-new DSI Bulk Option can accelerate loading for subscription material, or -- more commonly -- improve performance when a shop needs to replicate large batches of insert statements in a single table. Ditto for Replication Server's support for non-blocking commits, which leverages a feature inside Adaptive Server Enterprise (ASE, an embedded version of which ships with Replication Server) to boost replication performance; it's supported in Oracle 10g (version 2 or later).
Elsewhere on the performance front, Replication Server 15.2 can now replicate SQL statements instead of individual row changes. Zhang positions this feature as a complement to log-based replication.
On the management front, Sybase now includes a canned Replication Server environment that's designed to automate the installation and configuration process. Zhang says this will be of special interest to customers in non-ASE shops because it's designed to help them get quickly up and running.
Sybase has an enviable data integration (DI) product portfolio, encompassing data federation (Sybase Data Federation), ETL (Sybase ETL), data modeling (Sybase PowerDesigner), and other tools. Zhang says Sybase is focusing on identifying synergies between Replication Server and these other offerings.
"We are doing some investigation in terms of how to take advantage of replication as a technology [for] change data capture and then link in to Sybase ETL. We are thinking of ways of how to link the two," he says, adding that such an effort is "a forward-looking thing" that Sybase plans to deliver in upcoming iterations of the Replication Server and ETL products.
Synergistic relations between Replication Server and the (near-)ubiquitous PowerDesigner are more concrete, he claims. Sybase offers a pair of Replication Server packages: a basic package (which includes only Replication Server) and a premium package (which includes a copy of PowerDesigner and replication agents for Microsoft, Oracle, and IBM DBMSes).
"In PowerDesigner, we have a module called Data Architect. In Data Architect, you can use PowerDesigner to design replication server applications. In this design pattern, you would do reverse-engineering of the source database, reverse engineering of the target database, and then drag that replication server icon in the middle. Then you connect the dots and define what table you want to replicate from and what table you want to replicate to. This is one of the ways that we're actually helping customers improve their productivity through PowerDesigner. They don't have to manually create replication scripts," he explains, adding that this can help accelerate heterogeneous replication -- such as between Oracle and Microsoft databases.
Nor is this the extent of the PowerDesigner tie-in.
"A lot of customers are using PowerDesigner to reverse engineer their existing replication system, then suddenly all of that metadata will be in PowerDesigner, then suddenly you can say, 'I want to drop a column, I want to add a column. Not only can you do design work going forward … [but the] customer can also use PowerDesigner to conduct impact analysis," he explains, citing a scenario in which a customer uses PowerDesigner to model the upstream or downstream impact of adding or dropping columns from replication schemes.
Sybase has an ambitious year in store for Replication Server: version 15.2 delivers support for Oracle 11g; in subsequent releases -- this year -- Sybase plans to introduce support for "different [Oracle] database features, such as index organized tables, such as Oracle DataGuard [replication] technology." Sybase plans to release Oracle-themed updates for Replication Server on a per-quarter basis for the remainder of the year, Zhang concludes.