Häufig gestellte Fragen
What distinguishes hyperconverged infrastructure from a classic server-SAN architecture?
In a classic three-tier architecture, compute (servers), storage (such as a SAN) and network are built as separate hardware silos and managed separately. HCI merges these layers in software on a cluster of standard servers into a shared resource pool that is controlled through a single interface. For ERP environments, this generally means less administrative effort and simpler growth, because compute and storage are expanded together by adding nodes. A single, very performance-hungry ERP database, on the other hand, may still benefit from a dedicated SAN architecture depending on requirements, which is why the decision depends on the specific workload profile.
How many nodes does an HCI cluster need at minimum for resilient ERP operation?
For built-in redundancy, the common platforms usually require at least three nodes, so that the failure of a single server does not lead to data loss. With VMware vSAN, three hosts suffice for a fault tolerance of FTT=1 with RAID-1 mirroring; for FTT=2 with mirroring, at least five hosts are needed according to the 2n+1 formula (six are often recommended for recovery reserves); Nutanix requires at least three nodes for Replication Factor 2 and at least five for Replication Factor 3. For smaller sites, there are two-node variants with an external witness in a separate fault domain. In practice, an additional node (N+1) is often planned in, so that sufficient capacity remains available even during maintenance or recovery.
Is HCI certified and released for productive SAP HANA workloads?
Yes, both Nutanix and VMware vSAN hold SAP certifications for productive HANA operation on hyperconverged infrastructure. Nutanix states that it was the first vendor with a production release for SAP HANA on HCI (scale-up 2018, scale-out 2020), and for vSAN, SAP specifies a minimum of three and a maximum of 64 nodes per HANA HCI cluster. Importantly, for SAP HANA production workloads only the officially certified HCI solutions and certain minimum releases are supported. Before an implementation, the relevant SAP Notes and the current certification list should therefore be checked, since configuration limits differ depending on processor and version.
Does the built-in redundancy of HCI replace a dedicated backup and disaster recovery concept?
No. The distributed, redundant storage protects against the failure of individual hardware components within the cluster, but does not replace an independent data backup. Logical errors, accidental or malicious deletion, ransomware or the loss of an entire site are not covered by cluster redundancy alone. For this, HCI platforms offer native functions such as snapshots and replication, increasingly supplemented by immutable (WORM) backups; in addition, integrated backup solutions from third-party vendors can be connected. For a business-critical ERP, a tested backup concept and a disaster recovery strategy with defined recovery objectives are therefore mandatory.
For which companies is HCI worthwhile for ERP hosting — and for which is it not?
HCI is particularly interesting when a company continues to run its ERP itself (on-premises) or in its own private data centre and prioritises high availability, predictable growth and control over its own infrastructure. Since HCI scales compute and storage together and consolidates several workloads — such as ERP, database, document management and test systems — on one platform, it is often considered economically attractive for complete infrastructure renewals and mixed workloads. Companies that obtain their ERP as SaaS, however, run it at the provider's site and do not need their own HCI platform. For very small environments, the usual minimum size of three nodes plus software licences can make the investment disproportionate, which is why HCI should always be evaluated within the overall total cost of ownership.