🔍 Read the full analysis: How Idle SAP HANA Memory Can Add Capacity In Singapore Data Centers on Rymvard

Prime Big Deal Days · Oct 6–7Offer from Amazon

Get networking and server gear delivered free — and shop member deals

• Fast, free delivery on millions of items
• Access to Prime Big Deal Days deals on October 6–7
• Prime Video, Amazon Music and more included

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

TL;DR

Rymvard says its early-access software can identify SAP HANA memory allocations that have gone largely unused, helping hosting providers assess whether existing Singapore data-center capacity can be used more efficiently. The company illustrates the approach with a hypothetical regional hosting estate, not a customer deployment; no measured capacity gains or customer outcomes are disclosed.

Rymvard says its early-access software can help SAP hosting providers in Singapore identify memory reserved by SAP HANA systems but little used, potentially freeing existing server capacity without adding data-center floor space. The company’s example is an illustrative regional hosting estate, not a customer deployment, and no realized capacity gains are reported.

The product compares each HANA system’s allocation limit with its peak memory use over a rolling 30-day period. It flags hosts that are overbooked or above 90% utilization and identifies systems that used less than one-tenth of their allocation throughout the period, giving operators candidates to review for right-sizing. These flags are indicators for operational review, not evidence that every allocation can safely be reduced.

For each host, Rymvard separates physical memory into operating-system reserve, HANA systems’ observed peak use, allocated but unused memory, memory set aside for replicas at other sites, and genuinely free memory. Its default operating-system reserve is 10% of the first 64 GiB plus 3% of the remainder, unless a host provides a different reserve. Allocations above physical memory appear as overbooking rather than being presented as negative free memory.

Rymvard says it checks takeover requirements for each replicated system pair before recommending changes, so memory needed for recovery is not treated as available for release. The company describes the screens as coming from its running product, but says the estate shown is illustrative. It has not named a customer, supplied an independently verified result, or published a price; pricing is agreed with early-access partners.

Finding Capacity Inside Existing Servers

Singapore’s data-center supply is constrained by a selective allocation process, making the use of already installed infrastructure relevant to hosting providers. If a HANA system has held a large allocation while using only a small fraction of it, a carefully validated reduction could let operators fit more workloads within existing physical memory. That could defer some hardware expansion, though the extent of any benefit depends on each estate and has not been quantified by Rymvard.

The issue is not simply how much memory appears unused during ordinary operation. SAP workloads may need spare capacity during demand peaks or after a failure, and replicas can require memory at another site when a primary system is unavailable. Rymvard’s distinction between idle allocation, true free memory, and replica reserve addresses those separate uses. Its analysis may help operators see those trade-offs, but it does not replace workload testing, change controls, or recovery planning.

For customers, potential effects include more hosting capacity from current infrastructure or a more evidence-based decision about when additional servers are needed. There is no reported customer case establishing either outcome. Any capacity estimate should be treated as a planning opportunity until an operator validates it against actual workload behavior and recovery obligations.

Singapore’s Selective Data-Center Capacity

Singapore has used calls for applications to allocate new data-center capacity. In 2023, four proposals were selected under a pilot call, representing about 80 MW in total. A second call opened in December 2025. The capacity-allocation process is relevant to providers considering expansion, but it does not establish that all available capacity is exhausted or that memory optimization can substitute for new facilities.

Rymvard frames its product around a different resource: memory already installed in a hosting environment. HANA systems can have allocation limits substantially above their recent use, but the difference is not automatically safe to reclaim. Usage can vary, and standby systems may need resources during a recovery event. The product’s 30-day peak measure and separate accounting for replicas are intended to make those distinctions visible to operators.

Singapore banks’ technology risk requirements also shape the operational question. The supplied description cites MAS Notice 644 and a recovery time objective of no more than four hours for each critical system. Rymvard says it retains takeover checks for replicated pairs when identifying possible memory reductions. The notice’s requirements apply to regulated institutions; this description alone does not establish how a particular provider or customer meets them.

““Rymvard is in early access.””

— Rymvard

No Customer Capacity Gains Reported

Rymvard has not disclosed a named customer, a live customer estate, or a measured amount of memory or data-center capacity recovered through the product. Its example is explicitly illustrative, so it cannot establish how much capacity a provider in Singapore—or any specific operator—could release. The company also has not provided independent validation of the product’s recommendations or details on the number of systems and deployments assessed.

It is also unclear how operators would translate a flagged allocation into a safe production change. The description does not provide thresholds for workload-specific approval, the length of observation needed beyond the stated 30-day window, or results from recovery tests. The stated reserve calculation is a default, and actual host settings may differ. Pricing, partner numbers, and a broader release timetable have not been announced.

Early-Access Testing Will Matter

The next useful evidence would be results from early-access partners: how often the tool identifies genuinely reducible allocations, how much memory operators approve for release, and whether those changes remain safe under peak workload and recovery conditions. Rymvard has not announced a date for general availability or a formal launch milestone.

Operators considering the approach would need to review the utilization history, confirm system-specific memory requirements, and test replica takeover capacity before making changes. Until customer outcomes or independent evaluations are disclosed, the product’s Singapore example remains a description of a possible capacity-planning method rather than proof of added hosting capacity.

Primary source: IMDA · via Rymvard

Key Questions

What is Rymvard offering?

Rymvard says its early-access product analyzes SAP HANA memory allocations and recent use, then flags possible right-sizing candidates and distinguishes unused allocations from free memory and replica reserves.

Has the tool freed capacity for a Singapore customer?

No customer outcome is identified. Rymvard says the Singapore hosting scenario is illustrative and does not imply a customer, site, or result.

How does the product identify underused systems?

It compares allocation limits with each system’s peak use over 30 days and flags systems that used less than 10% of their allocation during that period. Operators would still need to validate any proposed reduction.

Does right-sizing remove memory needed for recovery?

Rymvard says it checks takeover needs for replicated system pairs and accounts separately for replica memory. No independent recovery-test results are provided.

When will the product be generally available, and what will it cost?

Rymvard describes the product as being in early access. It has not announced general availability or a standard price; pricing is agreed with early-access partners.

Primary source: IMDA · via Rymvard