HANA News Blog

SAP HANA & Transparent Huge Pages (THP

Matthias Sander • 26. Januar 2026

Switched to madvise in 2025 - now this?

If you’re running SAP HANA on SLES 15 SP5-SP7, keep an eye on this new SUSE article 22286:
👉 Transparent Huge Pages (THP) with madvise can trigger high swap usage, performance issues, or even system hangs on multi-NUMA systems.

Why this matters:
- SAP recommended switching to THP = madvise (SAP Note 2131662) for SLES15 SP5 / RHEL 9.2 and higher
- On multi-NUMA nodes, this can backfire
- One NUMA node runs low on memory -> direct reclaim & compaction kick in
- Result: swap fills, reclaim spikes, HANA stalls or failover


You can find out if THP is active by command:


cat /sys/kernel/mm/transparent_hugepage/enabled

If you see the following result, the THP is active.


[always] madvise never


If the OS version is

  • SLES 15 SP5 and
  • RHEL 9.2

or higher, then THP is recommended to be active.


Typical red flags:
- Swap usage grows while other NUMA nodes still have free memory
- High direct reclaim activity: pgscand/s spikes in sar -B output.
- Counters like allocstall_movable, pgscan_direct, compact_stall increase (/proc/vmstat)
- System hangs under load or cluster failover

Status:
- Kernel fix in progress (see SUSE TID 22286)
- SUSE can provide TEST / PTF kernels


On multi-NUMA node systems high swap usage, hangs or performance issues were reported when THP was enabled due to an OS kernel issue. An OS kernel is currently being developed. So far, it is recommended to deactivate THP defragmentation on OS level by executing:

echo "never" > /sys/kernel/mm/transparent_hugepage/defrag

echo 0 > /sys/kernel/mm/transparent_hugepage/khugepaged/defrag

📌 Not affected:
- Single-NUMA systems
- Older SLES versions with THP fully disabled

- RHEL (currently no feedback and proven)

Bottom line:
Check your NUMA balance before HANA does it the hard way.


Sources:

2131662 - Transparent Huge Pages (THP) on SAP HANA Servers
https://me.sap.com/notes/2131662/E
3523701 - Transparent Huge Pages (THP) alert occurs
https://me.sap.com/notes/3523701/E
2031375 - SAP HANA: Transparent HugePages (THP) setting on Linux
https://me.sap.com/notes/0002031375
2684254 - SAP HANA DB: Recommended OS settings for SLES 15 / SLES for SAP Applications 15
https://me.sap.com/notes/2684254
2578899 - SUSE Linux Enterprise Server 15: Installation Note
https://me.sap.com/notes/2578899
SAP HANA swap and system hangs with THP on multi-NUMA node systems
https://support.scc.suse.com/s/kb/SAP-HANA-swap-and-system-hangs-with-THP-on-multi-NUMA-node-systems?language=en_US


SAP HANA News by XLC

HANA performance degradation after upgrade to SPS07+SPS08

Ongoing HANA HEX performance issues with SPS07+SPS08

von Jens Gleichmann • 9. Januar 2026
With SPS06 and even stronger in SPS07 the HEX engine was pushed to be used more often. This results on the one hand side in easy scenario to perfect results with lower memory and CPU consumption ending up in faster response times. But in scenarios with FAE (for all entries) together with FDA (fast data access), it can result in bad performance. After some customers upgraded their first systems to SPS07 I recommended to wait for Rev. 73/74. But some started early with Rev. 71/72 and we had to troubleshoot many statement. If you have similar performance issues after the upgrade to SPS07 feel free to contact us! Our current recommendation is to use Rev. 74 with some workarounds. The performance degradation is extreme in systems like EWM and BW with high analytical workload.
SAP HANA NSE - a technical deepdive with Q&A

HANA NSE - technical details Q&A

von Jens Gleichmann • 24. November 2025
SAP NSE was introduced with HANA 2.0 SPS04 and based on a similar approach like data aging. Data aging based on a application level approach which has a side effect if you are using a lot of Z-coding. You have to use special BADI's to access the correct data. This means you have to adapt your coding if you are using it for Z-tables or using not SAP standard functions for accessing the data in your Z-coding. In this blog we will talk about the technical aspects in more detail.
Partitioning process

Golden Rules for HANA Partitioning

von Jens Gleichmann • 24. November 2025
SAP HANA scaling and tuning with proper partitioning designs
R+R: intersection and missing services

From Complexity to Clarity: Optimizing the R+R List for SAP Cloud ERP

von Jens Gleichmann • 25. September 2025
In transformation projects like SAP RISE / SAP Cloud ERP, the Roles & Responsibilities (R+R) list often serves as the backbone for collaboration between customer, partner, and SAP. Yet too often, this list is treated as a static document rather than a living framework. Sometimes nobody knows exactly what was defined by
Why Databases Need Optimizer Statistics – With a Focus on SAP HANA

SAP HANA data statistics for the optimizer

von Jens Gleichmann • 28. Mai 2025
In the world of modern database management systems, query performance is not just a matter of hardware—it’s about smart execution plans. At the core of generating these plans lies a critical component: optimizer statistics. This article explores why databases need optimizer statistics, with particular emphasis on SAP HANA, while drawing parallels with Oracle, Microsoft SQL Server (MSSQL), and IBM DB2.
HANA OS maintenance

HANA OS maintenance

von Jens Gleichmann • 27. Mai 2025
Please notice that when you want to run HANA 2.0 SPS07, you need defined OS levels. As you can see RHEL7 and SLES12 are not certified for SPS07. The SPS07 release of HANA is the basis for the S/4HANA release 2023 which is my recommended go-to release for the next years. Keep in mind that you have to go to SPS07 when you are running SPS06 because it will run out of maintenance end of 2023.
The HANA Optimizer is one of the core components of the HANA database. It determines how SQL is exec

SAP HANA Optimizer - the heart of the performance

von Jens Gleichmann • 25. Mai 2025
A database optimizer behaves similarly to the navigation system in your car. You use various parameters to determine which route you want to take to reach a particular destination. Potential additional costs for tolls and ferries also play a role, as do other limitations such as the vehicle's height, length, and width. From these input parameters, the navigation system determines the optimal route based on current traffic information (traffic volume, construction sites, congestion, accidents, closures, etc.), weather conditions, and the length of the route. This means that with exactly identical input parameters, different routes, costs, and thus different travel times can arise depending on the given conditions.
Is NSE worth the effort or is the better question: Do you know your cost per GB of RAM?

Is the effort for NSE worth it?

von Jens Gleichmann • 18. April 2025
Most of our presentations on data tiering projects end with these typical questions: How much we will save? How fast can it be implemented? Is the effort worth it over time? My counter question: "Do you know how much 1 GB of memory costs your company per month or year?" => how much memory we have to save to be beneficial?
Buch: SAP HANA Deepdive

Buchtipp: SAP HANA Deepdive: Optimierung und Stabilität im Betrieb

von Jens Gleichmann und Matthias Sander • 30. März 2025
Unser erster Buch mit dem Titel "SAP HANA Deepdive: Optimierung und Stabilität im Betrieb" ist erschienen.
More time to switch from BSoH to S/4HANA

SAP ERP Extended Maintenance until 2033 - but with conditions

von Jens Gleichmann • 7. Februar 2025
Recently handelsblatt released an article with a new SAP RISE option called SAP ERP, private edition, transition option. This option includes a extended maintenance until the end 2033. This means 3 years more compared to the original on-prem extended maintenance. This statement was confirmed by SAP on request of handelsblatt, but customers receive more details, such as the price, in the first half of the year. This is a quite unusual move of SAP without any official statement on the news page. Just to raise more doubts? Strategy? However a good move against the critics and the ever shorter timeline. Perhaps it is also a consequence of the growing shortage of experts for operating and migrating the large number of systems.
more