- Key takeaways
- 1C:Enterprise workload specifics
- Processors (CPU): Debunking the core-count myth
- Storage subsystem: IOPS matter more than capacity
- RAM: Capacity and channel width
- DBMS in 2026: PostgreSQL vs MS SQL
- Virtualization vs Bare Metal
- Practical sizing: Sample configurations
- Frequently asked questions (FAQ)
- Conclusion
We are ready to help you
1C and corporate databases in 2026: Step-by-step server selection guide for high loads
Key takeaways
- 1C architecture:Enterprise is critically dependent on single-thread performance; server CPUs should have a base clock of at least 3.6 GHz (ideally from 4.0 GHz in Turbo mode).
- Classic SATA SSD drives are completely unsuitable for databases over 100 GB; the 2026 standard requires NVMe PCIe 5.0 drives rated from 1,000,000 IOPS.
- Incorrect NUMA configuration on dual-processor servers reduces memory access speed by 30–40%, degrading the entire system.
- The corporate sector’s shift to PostgreSQL requires 20–30% more server RAM than comparable databases on MS SQL Server.
1. 1C:Enterprise workload specifics
1C:Enterprise is a classic example of a three-tier architecture (Client — Application Server — DBMS Server), where each tier needs specific hardware. The platform’s main trait is strict sequential (serial) processing of transactions when writing registers.
Amdahl’s Law is merciless in 1C: overall system speed is limited by single-thread data processing. A server with 64 cores at 2.2 GHz will run 1C twice as slow as a server with 8 cores at 4.5 GHz.
Waiting for a slow disk (I/O Wait) instantly creates transaction lock queues. If a lock lasts longer than 15 seconds, users get a timeout error and their sessions are terminated.
2. Processors (CPU): Debunking the core-count myth
Choose a CPU for 1C solely by base clock and L3 cache size. For a 1C application server in 2026, optimal choices are Intel Xeon Gold 65xx (for example, 6544Y) or AMD EPYC F-series (Frequency-optimized).
Many-core CPUs are needed only for a dedicated DBMS server, where each request from the 1C server can be processed in parallel. For the application server, buying a chip with more than 16 cores is wasted IT budget.
“The main mistake sysadmins make in 2026 is buying used servers with Xeon v4 CPUs at 2.0 GHz for ERP databases. 1C needs gigahertz. If core frequency is below 3.5 GHz, no SSD arrays or terabytes of RAM will save you from lock queues.”
NUMA (Non-Uniform Memory Access) architecture in dual-processor servers matters greatly. If the 1C server is misconfigured, a core on the first CPU will access the second CPU’s memory, adding up to 120 nanoseconds of hardware latency per operation.
3. Storage subsystem: IOPS matter more than capacity
A database storage subsystem is measured by IOPS (input/output operations per second) on random 4K reads. In 2026, standard SATA SSD drives capped at 90,000 IOPS became the main bottleneck for corporate databases.
The modern standard is NVMe U.3 drives with a PCIe 5.0 interface connected directly to CPU lanes. They deliver over 1,500,000 IOPS at latencies under 0.05 ms. PCIe 5.0 interface performance specifications on the official PCI-SIG portal
Comparison table: Drive performance for databases
| Drive type | Bandwidth | IOPS (Random 4K) | Latency |
|---|---|---|---|
| SAS HDD 10K (Legacy) | Up to 250 MB/s | ~150 - 200 | 5 - 10 ms |
| SATA SSD Enterprise | Up to 550 MB/s | ~90 000 | 1 - 2 ms |
| NVMe PCIe 4.0 | Up to 7 000 MB/s | ~800 000 | < 0.1 ms |
| NVMe PCIe 5.0 (2026 standard) | Up to 14 000 MB/s | > 1 500 000 | < 0.05 ms |
Array architecture requires physical separation of data streams. SQL data files (mdf), transaction logs (ldf), and the temp database (TempDB) should sit on physically separate RAID arrays (RAID 10 recommended) to avoid controller cache contention.
4. RAM: Capacity and channel width
DBMS server RAM must hold the entire active database plus 20% reserve for the OS. If your database is 300 GB, the SQL server needs at least 384 GB RAM DDR5 with ECC.
Memory bandwidth is critical for fast table retrieval. Modern CPUs support 8- or 12-channel memory. To reach maximum speed, install DIMMs symmetrically, filling all available CPU channels.
5. DBMS in 2026: PostgreSQL vs MS SQL
The corporate database market changed radically in 2026: over 70% of new 1C:ERP deployments run on PostgreSQL due to import substitution policy. Database choice directly dictates hardware requirements.
PostgreSQL uses more RAM with aggressive caching (shared_buffers setting) than MS SQL. Also, Autovacuum dead-row cleanup creates background disk load, making NVMe drives mandatory.
Pros & Cons (Switching to PostgreSQL for 1C)
No license fees (saving millions of rubles), official 1C platform support, open source, and sanction independence.
Requires highly skilled DBAs for fine tuning, high sensitivity to storage speed, non-standard query planner behavior on complex queries.
6. Virtualization vs Bare Metal
Using VMs (VMware, Hyper-V, KVM) for databases over 500 GB adds overhead. The hypervisor layer inevitably takes 10 to 15% of CPU clock for internal interrupt routing.
The main virtualization problem for databases is micro-latency in storage when requests pass through virtual switches.
For systems with more than 300 active users, the DBMS server must run on physical hardware (Bare Metal). Virtualization is justified only for 1C web publication servers and test environments.
7. Practical sizing: Sample configurations
Hardware sizing is based on concurrent users and document throughput. Below are standardized builds for the corporate sector in 2026.
Single-processor server. CPU 8 cores / 4.0 GHz. RAM 128 GB DDR5 ECC. Two NVMe SSDs at 1.92 TB in software mirror (RAID 1). Combined role (1C Server + DBMS).
Dual-processor server. Two CPUs with 16 cores at a base frequency of 3.6 GHz. RAM 512 GB DDR5. Hardware RAID controller with 4 GB cache and an array of 4 NVMe drives in RAID 10.
Cluster architecture. Three separate 1C application servers (high-frequency CPUs from 4.5 GHz). Dedicated fault-tolerant PostgreSQL DBMS cluster with replication and All-Flash SAN based on NVMe-oF (NVMe over Fabrics).
Conclusion
Infrastructure for 1C:Enterprise in 2026 tolerates no compromises like slow disks or low-frequency CPUs. Generic servers with dozens of weak cores can no longer deliver required transactional performance. Only a move to high-frequency architectures (from 4.0 GHz) and NVMe PCIe 5.0 arrays guarantees uninterrupted operation without lock queues and freezes.
Frequently asked questions We have prepared answers.
- How do you calculate required RAM for a 1C server? Allocate 2 to 4 GB RAM per 1C server worker process (rphost) depending on configuration. For the DBMS, RAM is calculated as 100% of the active database size plus 20–30% reserve for the OS.
- What if users complain about long locks when posting invoices? Move the TempDB database and transaction logs (ldf) to a separate ultra-fast NVMe array. Also check server BIOS power settings: power saving mode should be set to Maximum Performance.
- Why do consumer SSDs (Samsung PRO / Kingston) fail quickly in servers? Desktop SSDs have low endurance (DWPD — Drive Writes Per Day) and lack power-loss protection capacitors (PLP). Enterprise databases may only use Enterprise-class drives (Mixed Use or Write Intensive).
- Can you move the DBMS server to the cloud to save money? Cloud databases (DBaaS) work well for small business and startups. However, for heavy 1C databases (from 200 GB), network latency (Ping) between your office and the cloud data center will cause critical 1C UI slowdowns.