Elasticsearch 8.19.0, the open source search and analytics engine, deploys in one click onto a dedicated server Selfhost.dev provisions, secures and runs. The elastic password exists before the node first starts, so there is no generated password to catch in a terminal and security is on from the first request.
One container on a server only your project uses. Six locations on three continents.
Elasticsearch running on a server your project owns instead of inside Elastic's cloud. Your indices, mappings and the elastic account sit on that server's disk.
Developers adding search to an app, indexing logs from one service or analysing documents, who are comfortable sending JSON to a REST API.
When one node holds your data and a flat bill suits you better than a two zone cluster. Need replicas? Use Elastic Cloud or our Managed OpenSearch.
Elastic's own Docker Compose example runs three nodes plus Kibana and needs vm.max_map_count raised on the host. This template runs one node with the password, heap, memory lock and proxy settled before boot.
Elastic's docker run start generates the elastic password and prints it once in the terminal. Miss it and you reset it from inside the container. Here the password is written in before first boot, so the node never runs without it.
Your app, scripts and curl
TLS and routing
Elasticsearch 8.19.0
Indices and mappings
elastic password
| Elasticsearch on Docker Compose, by hand | Elasticsearch on Selfhost.dev |
|---|---|
Raise vm.max_map_count on the host or the node refuses to start | discovery.type=single-node, so the bootstrap checks do not apply |
Catch the generated elastic password or write one into .env | Set before first boot and shown in the console |
| Pick a heap size or let the JVM size itself | Fixed at 512 MB with ES_JAVA_OPTS |
Set bootstrap.memory_lock and the memlock ulimit together | Both set, so the heap is really locked |
| Put a proxy and a certificate in front of the node | TLS issued and renewed per hostname; port 9200 is never public |
| Pull an image tag and remember to move it | Pinned to elasticsearch-wolfi:8.19.0; a restart or redeploy brings back the same build |
Elasticsearch hosting requirements come down to the server size: one container with its indices on a volume behind TLS. We recommend 8 GB, with 512 MB for the heap and the rest as page cache, which is what makes search fast.
An Elasticsearch deployment with no package to install and no cluster to form.
This page opens the exact console template. Pick one of six locations and submit. The template deploys in about 10 minutes.
The console shows the elastic username and the password set before first boot. It is the only key to the endpoint, so store it safely.
Point curl or a client library at the HTTPS address, authenticate as elastic and PUT a document. The node answers in JSON.
Elasticsearch costs nothing to license under its open source terms, so hosting it here costs the server: 8 GB in Europe at $0.03/hr, about $23 a month, billed by the hour against prepaid credits. Queries and documents add nothing.
| Item | Cost |
|---|---|
| Queries, documents and indices | $0. Nothing is metered |
| Bandwidth | $0. Traffic is included in the server price |
| Optional daily backups | +20% of the server price |
| Server | RAM, cores, disk | Where | Rate | About a month |
|---|---|---|---|---|
cx33 (recommended) | 8 GB, 4 cores, 80 GB | Europe | $0.03/hr | $23.36 |
cax21 (ARM) | 8 GB, 4 cores, 80 GB | Europe | $0.04/hr | $29.20 |
ccx13 (dedicated cores) | 8 GB, 2 cores, 80 GB | US | $0.16/hr | $119.28 |
cpx32 | 8 GB, 4 cores, 160 GB | Singapore | $0.19/hr | $135.63 |
Monthly figures assume 730 hours. US and Singapore servers cost more for the same memory.
With a card on file or a past top-up, the server is powered off and held for 24 hours with an email; a top-up in that window restarts it. Welcome-credit-only servers are deleted at zero, so snapshot any index you cannot rebuild.
Put 8 GB in two locations into the calculator and compare, no signup.
Elastic Cloud Hosted starts at $99 a month: a minimum for a two zone production setup with 120 GB of storage. One node here is about $23. Same engine, different shape: theirs is replicated and ours is not.
| Elastic Cloud Hosted | Selfhost.dev | |
|---|---|---|
| Entry price | As low as $99 a month (Standard) | About $23 a month (cx33, Europe) |
| Shape at that price | Two zones, 120 GB of storage | One node, 80 GB disk |
| Replicas and failover | Yes, across zones | No |
| Who operates the engine | Elastic | You; we run the server |
Elastic list prices from elastic.co/pricing/cloud-hosted, each a minimum "based on cloud production config, 120 GB storage / 2 zones". Ours assumes 730 hours.
When a paid feature or a second zone is the requirement. SAML or OpenID Connect sign on, field and document level security, audit logging and machine learning are paid Elastic features wherever the software runs. If an index must survive a machine going away, buy the replicated cluster.
Elasticsearch's template deploys in about 10 minutes. That is the platform's own budget, not an estimate.
None of these a dealbreaker? The template deploys in about 10 minutes.
Launch ElasticsearchOptional daily server backups cover the whole server and go with it. Elasticsearch's snapshot API writes to a repository you register; those snapshots restore into any cluster, so they are your exit as well as your backup. S3, GCS and Azure repositories are built in; their keys go in the node's keystore, which takes container access over SSH, granted on request.
| Method | What it covers | Where it lives |
|---|---|---|
| Daily server backup (optional, +20%) | The whole server: the node, the volume, every index; the last 7 kept | With the server; deleted with it |
| Elasticsearch snapshot | The indices you choose, restorable into another cluster | The repository you register |
Snapshot your current cluster, register the same repository here and restore. A 9.x snapshot cannot restore into this 8.19 node; reindex from remote can pull those documents over HTTP once the old host is on the reindex allowlist.
Tell us how many documents you index and how big they are. We will reply with a server size, the monthly cost and whether one node is enough. One reply, from a person.
Yes. Since September 2024 Elasticsearch's source is available under AGPLv3, Elastic License 2.0 or SSPL 1.0; AGPLv3 is OSI approved. The free Basic tier includes authentication, role based access control and encrypted communications. What you pay for here is the server it runs on. See pricing.
On this template, 8 GB is the recommended server. The JVM heap is fixed at 512 MB and the rest of the memory works as page cache, which is what keeps searches fast. Elastic's advice is to give the heap no more than half the RAM. Here the heap stays at 512 MB on every server size, so a bigger server adds page cache, not heap.
A single node, deliberately. There are no replicas and no shard failover; discovery.type is set to single-node, which also skips the bootstrap checks a container cannot pass. For a high availability cluster across zones, Selfhost.dev's Managed OpenSearch runs two or more data nodes with dedicated manager nodes.
No. This template is the Elasticsearch REST API on its own, so the public URL returns JSON rather than a browser interface. If you want the Kibana UI, deploy the Elasticsearch and Kibana configuration instead, where Kibana is the public entrypoint and the raw API stays unrouted.
No. Elasticsearch is made by Elastic. AWS runs Amazon OpenSearch Service, built on OpenSearch, the fork it started from Elasticsearch 7.10 in 2021. This template runs Elastic's own Elasticsearch on a server in your Selfhost.dev project. The two have drifted apart since the fork, so match your client library to the engine.
Selfhost.dev runs managed PostgreSQL, MySQL, Redis, ClickHouse, OpenSearch and Kafka on AWS, plus Projects: a dedicated server we provision and operate, for apps from GitHub and for 44 one-click configurations across 38 apps. Pricing is pay as you go against prepaid credits, with no tiers and no subscriptions.
Best for: Elasticsearch, Elasticsearch with Kibana, Grafana, Metabase and managed PostgreSQL, MySQL, Redis, ClickHouse, OpenSearch and Kafka.
