Best for dev/test
Hourly
1.5× daily rate
- Evaluate cache capacity before committing to monthly
- Run load tests to determine optimal memory size
- No minimum term — provision and destroy freely
NubexCloud Memcached is a fully managed, in-memory Key-Value caching service that absorbs hotspot database reads before they happen. Sub-millisecond response times, 100,000 QPS per node, 100% Memcached protocol compatible — deploy in minutes, connect without changing a line of code.
Memcached exists for one purpose: to answer requests before they reach your database. Every spec below is built around that goal.
Managed Memcached handles the infrastructure — you configure your cache keys and let the service absorb your hotspot traffic.
Cache the data your application reads most frequently — user sessions, product catalogues, rendered page fragments, API responses, and counter values. Serve these from memory and eliminate the corresponding database round-trips entirely. Backend read load drops dramatically on first deployment.
When a server failure occurs, the service recovers automatically and clients reconnect immediately. High-availability mode ensures the cache layer never becomes a single point of failure for your application stack. Note: cache data is not persisted — after a failover event, the cache repopulates from the database on first access.
Comprehensive data statistics for real-time monitoring and historical analysis. Track cache hit rates, memory usage, eviction rates, and connection counts from the console. Configure alerts when hit rate drops below threshold — a leading indicator that your cache needs more capacity.
Expand cache capacity from the console with a single click — no migration, no application changes, no reconfiguration. Scale based on actual memory usage and hit rate data. Note: expansion requires a restart and clears the cache, so plan expansions during low-traffic windows.
Memcached instances are accessible only from within your VPC — completely isolated from the public internet. No public IP, no external exposure. Your cached data is protected by network-level isolation, and access is restricted to cloud hosts in the same region and subnet.
Fully compatible with the Memcached protocol — all standard Memcached commands (get, set, delete, add, replace, increment, decrement, flush_all) and stats commands work without modification. Migrate from self-hosted Memcache by updating your connection string and nothing else.
Every repeat database read for the same data is wasted compute. Memcached intercepts those reads at the cache layer — serving the answer from memory in microseconds, never touching the database again until the cache entry expires or is evicted.
Pick the memory capacity your hotspot dataset requires. Monitor hit rates and memory usage from the console — expand when the data tells you to.
| Memory Capacity | Max QPS | Key Size Limit | Value Size Limit | From / mo |
|---|---|---|---|---|
| 1 GB | 100,000 | 250 bytes | 1 MB | $6 |
| 2 GB | 100,000 | 250 bytes | 1 MB | $12 |
| 4 GB | 100,000 | 250 bytes | 1 MB | $24 |
| 8 GB | 100,000 | 250 bytes | 1 MB | $48 |
| 16 GB | 100,000 | 250 bytes | 1 MB | $96 |
| 32 GB | 100,000 | 250 bytes | 1 MB | $192 |
Prices in USD, estimated. Exact pricing via console. Billed by actual usage hour.
Both are in-memory caches. The right choice depends on what your data structure and durability requirements actually are.
| Capability | NubexCloud Memcached | Redis (see NubexCloud Redis) |
|---|---|---|
| Data model | Key-Value strings only | Strings, lists, sets, hashes, sorted sets |
| Pure caching performance | ✓ Fastest for simple K-V get/set | Slightly more overhead |
| Multi-threading | ✓ Multi-threaded — scales CPU cores | Single-threaded (Redis 6+ I/O threads) |
| Data persistence | ✗ Memory only — data lost on restart | ✓ RDB + AOF persistence options |
| Data structures | ✗ Strings only | ✓ Rich data structures |
| Pub/Sub messaging | ✗ Not supported | ✓ Native pub/sub |
| Atomic counters | ✓ incr/decr commands | ✓ INCR/DECR + more |
| Protocol compatibility | ✓ 100% Memcached protocol | ✓ 100% Redis protocol |
| Best use case | Pure cache · <100K QPS · Simple K-V | Session store · Queues · Leaderboards |
Memcached is purpose-built for high-frequency read workloads. These are the scenarios where deploying it pays off immediately.
Memcached billing is straightforward — memory capacity × duration. No per-request fees, no data transfer charges within the same VPC.
1.5× daily rate
Standard monthly rate
Pay 10 months, get 12
Memcached stores data in RAM only. Data is lost after a failover event or during a capacity expansion restart. Plan expansions during low-traffic periods. For workloads that require persistence across restarts, use NubexCloud Redis instead.
Sub-millisecond cache hits. 100K QPS. 100% Memcached compatible. Deploy in minutes — connect your existing application without changing a single line of code.