If I told you that the latest release of ASE, version 16 Service Pack 03, can do a million plus database transactions a minute by scaling linearly on 64 cores without much tuning, or that, the performance is 40X better than what you currently have with ASE 15.7 without any application changes, or that, ASE has smart in-memory technology that makes optimal use of memory while being governed by a customer-defined data lifecycle policy, you may say “no way!” or “show me!”.
Well, all of this is true! If you take a deeper look at ASE 16 SP03, you will find some new exciting capabilities in the MemScale option of ASE that can inject more power into your high performance, mission critical OLTP systems:
- In-memory row store (IMRS) with data row caching (DRC). ASE now supports an IMRS for ACID-compliant OLTP processing by using DRC to store frequently accessed rows of a table. Note that this is frequently accessed rows only and not the whole table plus the fact that ASE automatically packs in-memory rows that become less frequently used back to the page store/buffer cache.
- New indexing schemes. This release supports Hash Cached B-tree (HCB) indexing, which combines the advantages of traditional B-tree structures with hashing. B-tree indexing continues to be supported for data in traditional page stores, but the active portions of a B-tree index is now cached in a hash index structure.
- Multi-version concurrency control (MVCC). To further improve concurrency in the system, ASE 16 supports MVCC both on-disk as well as for IMRS. With MVCC, readers and writers do not block each other.
So, what is the benefit of all these features? Up to 40 times better performance. By storing hot data in IMRS and further avoiding the overheard of repeated compression and decompression, throughput is significantly higher, particularly when there is higher concurrency with high core-count systems. HCBs can improve query execution speeds for frequently accessed rows in DRC by reducing logical reads that would be incurred with traditional B-tree traversals and by faster join processing. Lab tests also show several times better performance for concurrent updates on hot pages.
How to take advantage of this new in-memory technology in ASE? The challenge is that users may not know how much memory to allocate for IMRS, how much space to allocate for the IMRS log or what the IMRS cache size should be for some production workload. No need to fret over this – ASE helps you decide what configuration changes are recommended to move to IMRS store and which applications are suitable to have in-memory technology enabled. A new tool – Workload Profiler – enables ASE to automatically recommend configuration changes to move from 100% page-based store to having hot rows in memory that leads to significant improvement in performance. The Workload Profiler tool joins the ranks of Workload Analyzer, made available with an earlier release to capture a production workload and replay on a new test server. With this release, the Workload Analyzer and Workload Profiler can be used in a complementary fashion to achieve improved performance without manual intervention or tuning. Lab tests have shown over 60% performance gains on TPCC-like workloads with in-memory processing without the hassle of coming up with optimal configuration values through trial and error.
Support for MVCC should significantly enhance performance of reporting application that you run on ASE. If these enhancements are still not sufficient for reporting and complex queries, then you can bring the power of SAP HANA to electrify the performance you get from ASE. SAP HANA is a fully in-memory, column-oriented RDBMS, which is integrated with ASE to meet the demanding needs of business. When HANA is bolted to ASE, queries directed at a reporting ASE server are pushed down and executed on HANA and results returned to the user. With minimal application changes and no other manual intervention, reporting applications can see almost 100 times improvement in performance!
Why stop at performance? With this release, ASE is supporting building new applications with SAP HANA’s SQL dialect called SQLScript. In this first phase, ASE supports SQL syntax, such as datatypes, DML, functions, operators, etc., procedural logic and features needed to build OLTP applications. Such applications need to use its own database, separate from databases supporting TSQL applications, under the management of a single ASE server. With this support, an OLTP application built on ASE can be run on HANA without any changes and vice-versa, thereby providing customers the flexibility to build once and deploy anywhere you run ASE or HANA. Subsequent phases of support in future releases will ensure full support of the language to provide flexibility and power to build new applications.
Finally, significant moves to get ASE to the cloud. With this release, ASE is supporting backup to AWS cloud using Amazon’s FileGateway solution. ASE is also available on AWS, Azure and Google Cloud Platform (planned), where you can run ASE or Business Suite on ASE using Bring-Your-Own-License (BYOL) model. Watch this space for more cloud related announcements on new data management services with ASE we plan to support in the future.
Want to learn more? Please register for the upcoming series of webinars below on the exciting new features in ASE and SAP Replication Server 16 SP03 at: https://event.on24.com/wcc/r/1480117/EC65E8FCB99D96463A86466A9157D459/150696
|August 22, 2017||What is New with SAP ASE 16 in 2017 and Beyond|
|Sept 05, 2017||MemScale Part 1: Technical Overview of xOLTP Features in ASE 160 SP03|
|Sept 12, 2017||MemScale Part 2: Multi-Version Concurrency Control, Migration Considerations|
|Oct 03, 2017||SAP ASE 16 Cloud Strategy and Offerings Available Today|
|Oct 09, 2017||The Debut of SAP Replication Server 16 and its Future Direction|
|Oct 24, 2017||The value proposition of SAP HANA Accelerator for ASE 2.0 SP02|
|Nov 07, 2017||SAP ASE Workload Analyzer Option 16 SP03 for upgrade, migrations and diagnostics|
For additional information, please visit the ASE community page at: https://www.sap.com/community/topic/ase.html