Cloud-based debut of SAP's HANA platform
SAP's HANA Enterprise Cloud service, in close collaboration with Intel, has shifted its focus towards commodity hardware for its cloud service, moving away from non-commodity processors like Graphics Processing Units (GPUs). This decision comes as part of the HARNESS project, led by Professor Alexander Wolf of Imperial College London's computing department, which aims to develop technologies for in-memory computing in the cloud.
The HANA Enterprise Cloud service, hosted by SAP and its managed service provider partners, runs on infrastructure built using Intel's Xeon Processor E7 family of chips. Each chip in this family boasts ten processor cores, designed for highly scalable workloads. Diane Bryant, general manager for Intel's data centers group, has stated that this collaboration will provide customers with the opportunity to deploy mission-critical solutions powered by SAP HANA with cloud simplicity.
To maintain performance in the cloud, HANA Enterprise Cloud employs several key strategies. In-memory storage, where data is stored primarily in main memory (RAM), allows for tens of thousands of times faster access for write operations compared to traditional storage. The service also supports both scale-up (more resources on single nodes) and scale-out (adding nodes in distributed systems) architectures to efficiently handle growing workloads and user concurrency, ensuring consistent write performance even under heavy loads.
Real-time data replication minimises latency by continuously synchronizing data with source systems, enabling immediate access to fresh data without performance degradation during write operations. Hybrid Transactional/Analytical Processing (HTAP) allows SAP HANA to perform online transactional processing (OLTP) and analytical processing (OLAP) on the same dataset concurrently, optimising both write and read workloads without bottlenecks.
Data lifecycle management and storage tiering are crucial in cloud environments. SAP HANA employs data aging and Native Storage Extension (NSE) to place less frequently written data on disk-based warm or cold storage layers. This reduces the in-memory footprint, allowing better memory resource allocation to active datasets with frequent write operations, thus maintaining performance.
Lastly, SAP's cloud provides observability tools that track CPU, memory, disk I/O, network traffic, and replication latency to optimise resource usage dynamically and prevent performance degradation during heavy write loads.
In-memory databases can be fast for read-only applications, but performance can degrade when data must be written to the database. However, SAP HANA Enterprise Cloud's in-memory, scalable architecture combined with cloud-native monitoring and tiered storage strategies ensures high-performance database write operations, with low latency and high throughput in writes within the cloud environment.
The new cloud-based version of SAP's HANA in-memory database platform will support SAP's ERP, CRM applications, and data warehouse offering. The service will provide customers with the opportunity to leverage the power of SAP HANA in a cloud environment, offering the benefits of cloud simplicity and scalability.
Data-and-cloud-computing technologies, such as SAP HANA Enterprise Cloud, are leveraging Intel's Xeon Processor E7 family of chips for their cloud services, ensuring high scalability for workloads. In the HARNESS project, led by Professor Alexander Wolf, technologies for in-memory computing in the cloud are being developed to maintain performance and ensure write operations are tens of thousands of times faster than traditional storage methods.
