Technology Blogs by Members
Explore a vibrant mix of technical expertise, industry insights, and tech buzz in member blogs covering SAP products, technology, and events. Get in the mix!
cancel
Showing results for 
Search instead for 
Did you mean: 
Rohit-Sharma
Participant
In this Article, I am going to explain following points for SAP HANA and Persistent memory

  • Persistent memory introduction

  • Advantage of Persistent Memory in SAP HANA infrastructure

  • Comparison of DRAM vs ( DRAM + PMEM ) infrastructure

  • SAP HANA Fast restart option and comparison with SAP HANA Persistent Memory

  • SAP HANA Hardware sizing with persistent memory  – Greenfield and Brownfield


Introduction of Persistent memory.

SAP HANA is an in-memory-first database. This means SAP HANA performs all database operations directly on the in-memory data structures.

Persistent memory is new type of memory with following main features. Persistent memory is supported with SAP HANA 2.00.035 and higher.


Persistent  Memory Introduction


Persistent memory (non-volatile RAM, NVRAM, virtual persistent memory, vPMEM, also referred to as storage class memory) is an additional persistent storage type that combines fast accesses (like memory) with durable storage (like disk) that isn't lost in case of crashes or restarts.

Below diagram explains different SAP HANA areas when moving from DRAM to DRAM+ PMEM (mixed) hardware infrastructure. SAP HANA controls what is placed in Persistent Memory and what remains in DRAM


Let’s understand both the options from below tables comparison.



















Option- A ( DRAM ) Option- B ( DRAM + PMEM)
DRAM hardware contains: SAP HANA work area + Column Store Tables + Row Store Tables PMEM contains Column Store tables while DRAM contains others segments as per above diagram
When the infrastructure includes only DRAM infrastructure, the initial table load occurs each time the database is started after planned or unplanned outages When the infrastructure includes persistent memory, the initial table load is no longer necessary. Data stored in persistent memory is retained across planned outage events and unplanned events, such as natural disasters.
In case of any power outage and unplanned events we can consider approx. 30-40 min additional time (based upon customer experience) for column store tables load in the SAP memory for large SAP HANA database No additional time for column store tables in case of any power outage and unplanned events.

Comparison between Intel® Optane™ DC Persistent Memory and Other Memory Types:
















































Memory Type Tier Memory Technology Non-Volatile Capacity TCO Restart time including data reloading
DRAM Hot DIMM No Low High Slow
Intel® Optane™ DC Persistent Memory Hot 3D Xpoint Yes High Medium Fast
SSD Warm NAND flash-based Yes High Low NA
HDD Cold Spinning disk-based Yes High Very low NA

Intel® Optane™ DC Persistent Memory is featured with:



    • High capacity: up to 512GB per memory slot (4x greater than traditional 128 GB DDR4 DIMM sizes), letting enterprises keep more data in the hot tier without complex deployment (scale out with multiple SAP HANA nodes, data tiering strategies, etc.)

    • Low TCO: lower costs for simplified infrastructure (update servers to support PMEM and scale-up instead of scale-out) and simplified data-tiering management (more data in hot and warm tiers)

    • Good access speed: near-DDR4 DRAM performance (similar read performance but slower write performance) in the same dual in-line memory module (DIMM) slot as DRAM.

    • Fast restart time: byte-addressable like DDR4 DRAM, but providing the same persistence as flash- and disk-based block storage that data is retained across server reboots and power outages.




As we are taking about SAP HANA persistent memory, so it is really important to mention few lines about SAP HANA fast restart option.

  • With SAP HANA SPS 2.00.040, SAP HANA has great feature : SAP HANA Fast restart option

  • Customers on DRAM infrastructure with SAP HANA 2.0 SPS 04 onwards, can go with SAP HANA Fast restart option:

  • The SAP HANA Fast Restart option uses storage in the file system to preserve and reuse MAIN data fragments to speed up SAP HANA restarts. This is effective in cases where the operating system is not restarted.


The fast restart option works in a similar way like persistent memory with the difference that the main storages are lost when the server is crashed or rebooted, however in case of SAP HANA restart column tables load will not happen with each restart .A quick comparison will help you to understand both options

















































Area Persistent Memory Fast Restart Option
SAP HANA Revision level >= 2.00.035 >= 2.00.040
RAM non-volatile RAM DRAM
Limited by persistent_memory_global_allocation_limit global_allocation_limit (>= 2.00.050)
persistent_memory_global_allocation_limit
Memory kept across SAP HANA restarts yes yes
Memory kept across OS restarts yes no
Infrastructure requirements limited (SAP Note 2618154) generally available

Now let’s understand SAP HANA Hardware sizing with Persistent memory :

SAP HANA Hardware Sizing with Persistent Memory:

SAP HANA with persistent memory follows the same sizing rules as SAP HANA with only DRAM

For Greenfield Sizing ( using quick sizer ) :

Greenfield sizing with Quick Sizer has a separate part for Persisted DB Memory


In the example, 237 GB can be placed in PMEM.


 

For Brownfield Sizing:

SAP HANA sizing report as of version 77 has an additional section on persistent memory

Sizing report provides NVRAM/PMEM and DRAM memory requirement.


In the example, 4844 GB can be placed in PMEM and 5374 GB in DRAM.

Kindly keep in mind while doing Persistent Memory sizing: (Refer SAP Note  2786237- Sizing SAP HANA with Persistent Memory )

  • Ratio between DRAM and PMEM capacity of 2:1, 1:1, 1:2 or 1:4.

  • Larger ratios only in exceptional cases (data tiering, where large amounts of warm data will be stored in PMEM).

  • Total capacity of up to 24 TB per physical server.

  • There should be sufficient hardware resources for the intended use case: Especially DRAM, persistent memory, and CPU.


Important References:

You can refer following note and articles on Persistent memory topic for more details.

 

Hope this article will help you to understand important points for Persistent Memory and SAP HANA Hardware sizing for Persistent memory for Greenfield and Brownfield implementation ,SAP HANA fast restart option comparison with Persistent memory.

Thanks !!

 

Kindly share feedback or thoughts in a comment or ask questions if any.

 

 
4 Comments
Labels in this area