🔍 Read the full analysis: Can Unused HANA Memory Help Ease Singapore’s Data Center Power Constraints? on Rymvard
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Rymvard, an early-access software provider, says its running product can help SAP hosting operators identify HANA memory allocations that have gone largely unused and right-size them. The example is illustrative, not a reported customer deployment, and the tool does not add data-center power or capacity; its potential benefit depends on verified workload data and recovery requirements.
Rymvard says its early-access software can help SAP hosting providers in Singapore identify HANA memory that has been allocated but used little or not at all, potentially improving use of existing data-center resources. The company’s example describes a regional hosting hub scenario, not a named customer, operating site or measured outcome, and the tool does not itself create new power or data-center capacity.
The product compares each HANA system’s configured memory allocation limit with its peak use over a rolling 30-day period, according to Rymvard. It flags hosts that are overbooked or above 90 percent utilization, and identifies systems that used less than one-tenth of their allocation throughout the 30 days for possible right-sizing. The measures are intended to show where memory might be reassigned rather than simply treating allocated memory as occupied.
Rymvard’s host-level view divides physical memory among the operating-system reserve, HANA systems’ recorded peak use, allocated memory that is unused, memory reserved for replicas at other sites, and memory that is genuinely free. The default operating-system reserve is 10 percent of the first 64 GiB and 3 percent of the remainder, unless the host specifies its own reserve. If system allocations exceed physical memory, the tool displays that as overbooking rather than masking it as negative free memory.
Recovery capacity is part of the proposed checks. Rymvard says each replicated system pair retains a takeover check during right-sizing, so memory needed for recovery is not treated as spare. That safeguard is relevant to Singapore banks, which under MAS Notice 644 must set a recovery time objective of no more than four hours for each critical system. The notice establishes the recovery requirement; it does not certify Rymvard’s product or validate a particular memory plan.
Can Unused HANA Memory Help Ease Singapore’s Data Center Power Constraints?
Rymvard says its running software can help SAP hosting operators spot HANA memory allocations that have seen little use and assess them for right-sizing. The example is illustrative: it reports no customer deployment, measured savings, or new data-center power.
Find capacity inside the estate
Rymvard’s early-access tool compares each HANA system’s configured memory allocation limit with its peak use over a rolling 30-day period. It flags overbooked or highly utilized hosts and highlights systems using less than one-tenth of their allocation for possible review.
Compare configured and used
Peak demand is measured against each system’s allocation limit, helping operators see where capacity may be consistently underused.
Keep reserves visible
The host view distinguishes recorded peaks, unused allocations, replica reserves, the operating-system reserve, and genuinely free memory.
Right-size with evidence
Low observed use is a prompt for workload and recovery checks. It is not proof that an allocation can safely be reduced.
Allocated does not always mean occupied
Rymvard’s host-level view aims to show where physical memory sits across active use, allocation headroom, recovery needs, and free capacity. The illustration below is conceptual, not a customer measurement.
The default OS reserve is 10% of the first 64 GiB and 3% of the remainder, unless the host specifies its own. If system allocations exceed physical memory, the tool displays overbooking rather than negative free memory.
Measure
Compare 30-day peaks with configured limits.
Flag
Identify highly utilized hosts and low-use systems.
Check recovery
Retain takeover checks for replicated system pairs.
Decide
Review workload history before changing allocations.
Efficiency can help, but it is not power supply
Singapore manages new data-center capacity through selective application rounds. IMDA said four proposals were selected in a 2023 pilot call, representing about 80 megawatts in total; a second call opened in December 2025. That backdrop makes better use of existing infrastructure relevant to operators.
What memory reuse might do
Reclaiming unused HANA allocation could help an operator fit workloads within existing server memory or avoid unnecessary expansion, if workload and recovery reviews support the change.
What it does not establish
Memory efficiency does not demonstrate spare electricity, cooling, suitable hardware, permission to expand, or reduced energy use. Rymvard has not quantified a power benefit.
Recovery needs stay in the calculation
Rymvard says each replicated system pair retains a takeover check during right-sizing, so memory needed for recovery is not treated as spare. That matters for Singapore banks: MAS Notice 644 requires a recovery time objective of no more than four hours for each critical system. The notice sets a recovery requirement; it does not certify this product or validate any particular memory plan.
| Question | What is known | What remains unverified |
|---|---|---|
| Does it identify low use? | Rymvard says it compares allocation limits with peak use over 30 days and flags systems below 10% use. | Whether that window captures seasonal peaks and rare business events at a given operator. |
| Is right-sizing safe? | Rymvard says takeover checks remain in place for replicated pairs. | Independent audits, customer validation, and performance or recovery outcomes. |
| Are savings proven? | The example describes an illustrative estate, not a named customer or site. | Reclaimed memory, electricity savings, deferred expansion, and measured results. |
| What are the terms? | The product is described as running in early access. | Public pricing and a deployment timetable; terms are agreed with participating partners. |
Real deployments would answer the hard questions
Useful evidence would show how much allocation operators safely reduced, which workloads were affected, and whether they reused the capacity without performance or recovery problems. Any electricity claim would also need a clear baseline and measurement period; memory figures alone cannot establish power savings.
What does the tool do?
It compares HANA allocation limits with 30-day peak use, then flags highly utilized hosts and systems below 10% use for review.
Does it add power or capacity?
No. It may help identify reclaimable allocations within existing systems. Actual capacity or power effects depend on deployment decisions and measured workloads.
Has a Singapore customer shown savings?
No customer outcome or quantified savings has been provided for the example. The scenarios are illustrative and imply no specific customer, site, or result.
Where Memory Reuse Could Help
Singapore manages new data-center capacity through selective application rounds rather than allowing operators to add capacity without constraint. The Infocomm Media Development Authority said four proposals were selected in a 2023 pilot Data Centre Call for Application, representing about 80 megawatts in total; a second call opened in December 2025. Against that backdrop, finding usable capacity inside existing facilities could be relevant to hosting providers that cannot quickly add another data hall.
But memory efficiency and power availability are related only indirectly. Reclaiming unused HANA allocation may let an operator fit additional workloads within a server’s memory resources or avoid unnecessary expansion. It does not establish that the operator has spare electrical supply, cooling capacity, suitable hardware, or permission to expand. Any power benefit would depend on whether memory right-sizing changes the number of servers required or their operating load, which Rymvard has not quantified.
For customers, the practical issue is whether freed capacity can be used without compromising system performance or disaster recovery. The dashboard’s comparison of allocation, observed peak demand and replica reserves could support that assessment. It is a planning aid, however, not evidence that every low-use allocation is safe to reduce.
Products Worth Considering
Anker SOLIX C1000 Gen 2 Portable Power Station
Anker SOLIX S2000 Portable Power Station, 2,010Wh, 1,500W
DJI Mic 3 Wireless Microphone System (2 Transmitters, 1 Receiver, Charging Case)
As an affiliate, we earn on qualifying purchases.
Singapore Capacity and HANA Reserves
Singapore’s data-center development has been shaped by controlled allocation of new capacity. The 2023 pilot selected four proposals for about 80 MW in all, and the government’s second application call opened in December 2025. This policy context makes operational efficiency one possible route for providers seeking to serve customers while new capacity is allocated selectively.
SAP HANA systems use memory-intensive in-memory data processing, and operators may configure allocation limits above recent usage to accommodate peaks or future needs. A low 30-day peak can be a signal to investigate, but it does not by itself establish that the allocation is unnecessary: usage could be seasonal, a rare business event may create a higher demand, or a recovery replica may need reserved headroom. Rymvard’s example therefore separates active use, unused allocation, replica requirements and free memory rather than presenting all unconsumed allocation as immediately available.
The product is described as running in early access. Rymvard says the screens are from the running software but use an illustrative example estate, not a customer environment. Its scenarios imply no specific customer, site or achieved savings. The company has not published prices, saying terms are agreed with early-access partners.
“The screens shown come from the running product on an illustrative example estate, not from a customer.”
— Rymvard
Savings and Safety Still Unverified
No customer results or quantified savings have been provided for the example. It remains unclear how much memory a real operator could reclaim, whether the change would reduce electricity use, or whether it could defer a server or facility expansion. Rymvard has not published pricing, and its early-access status means availability and commercial terms are being handled with participating partners.
The 30-day measurement window also leaves questions for deployment. Operators would need to determine whether it captures seasonal peaks and exceptional events, how allocation changes are approved, and how recovery tests are carried out. Rymvard says takeover checks remain in place for replicated pairs, but no independent audit, customer validation or performance data is cited here. A low observed peak should not be treated as proof that a system’s reserve can be cut without review.
Early Access and Real-World Testing
The next meaningful evidence would come from customer deployments showing how much allocation was safely reduced, what workloads were affected, and whether capacity could be reused without performance or recovery problems. For any claimed power benefit, operators would also need to report an appropriate baseline and measurement period, rather than infer electricity savings from memory figures alone.
Rymvard says interested organizations can discuss early-access participation directly with the company; it has not announced a public price list or a deployment timetable. Singapore’s second data-center application round is a separate policy development and does not establish whether providers using the software will receive additional capacity. Until customer outcomes are reported, the proposal remains a potential way to manage existing infrastructure more closely, not a demonstrated solution to the city-state’s power constraints.
Primary source: IMDA · via Rymvard
Key Questions
What does Rymvard’s HANA tool do?
Rymvard says it compares each HANA system’s allocation limit with peak use over 30 days, then flags overbooked or highly utilized hosts and systems using less than 10 percent of their allocation for possible review.
Does the tool add data-center power or capacity?
No. It identifies allocations that may be reclaimable within existing systems. Any effect on power use or the need for additional servers would depend on actual workloads and deployment decisions; Rymvard has not reported quantified savings.
Has a Singapore customer demonstrated savings?
Not in the information provided. Rymvard says its screens use an illustrative example estate, not a customer, and makes no claim of a measured customer outcome.
How does the approach account for disaster recovery?
Rymvard says it preserves takeover checks for replicated system pairs and separately identifies memory reserved for replicas at other sites. Singapore banks’ critical systems must have a recovery time objective of no more than four hours under MAS Notice 644, but that requirement does not independently validate the product.
When did Singapore open its latest data-center application call?
The second Data Centre Call for Application opened in December 2025. In the 2023 pilot, four proposals were selected for about 80 MW in total.
