Softwr

Technology · head to head

etcd vs Thought Machine

etcd logo

etcd

Technology

A distributed key-value store using Raft consensus, built for cluster coordination rather than application data.

From
On request
Rated
-
Full reviewWebsite
Thought Machine logo

Thought Machine

Technology

Cloud native core banking where products are written as smart contracts

From
On request
Rated
-
Full reviewWebsite

The short version

  • Each has a real cost: etcd it is sized for coordination data, not application data: the default backend quota is 2 GB and 8 GB is the documented recommended maximum, and exceeding it puts the cluster into a NOSPACE alarm where it accepts no writes until an operator compacts, defragments and clears the alarm by hand.; Thought Machine a core replacement is a two to four year programme at an established bank, and the licence is a minority of total cost against system integrator fees, parallel running and data migration.
  • They diverge on capability: etcd covers Raft consensus, Thought Machine covers Smart contract product engine.
  • Prices and features above were last checked on 1 September 2026.

Where they differ

Only the attributes on which etcd and Thought Machine actually diverge.

Attributes where etcd and Thought Machine differ
AttributeetcdThought Machine
Pricing modelopen-sourcequote
PlatformsWebWeb, API, Cloud

Identical on both: starting price (On request), free tier (No), user rating (Not yet rated), category (Technology).

What each one covers

Drawn from each product's published feature list. An absence here means we hold no record of it - not that the product lacks it.

Only in etcd

  • Raft consensus
  • Linearizable reads
  • Transactions
  • Leases
  • Watches
  • MVCC revision history
  • Role-based access control
  • Snapshot backup and restore

Only in Thought Machine

  • Smart contract product engine
  • Cloud native architecture
  • Real time ledger
  • Vault Payments
  • API first design
  • Multi entity and multi currency
  • Product versioning and testing
  • Configurable posting rules

What people use each for

The jobs each tool is most often brought in to do.

etcd

  • Storing Kubernetes cluster state, which is what the overwhelming majority of etcd deployments are doingnot Thought Machine
  • Leader election and distributed locking in a home-grown scheduler or control plane, using leases and transactionsnot Thought Machine
  • Service discovery and dynamic configuration where readers need to be notified of changes rather than poll for themnot Thought Machine
  • Coordinating failover in a clustered database, as Patroni does for PostgreSQLnot Thought Machine

Thought Machine

  • A large bank launching a separate digital brand on a clean core rather than migrating the back booknot etcd
  • A bank whose product launches are blocked by vendor change requests on a legacy corenot etcd
  • An institution needing real time balances and postings for instant payment obligationsnot etcd
  • A group consolidating multiple country cores onto one multi entity platformnot etcd

Where each one falls short

Documented limitations, not opinions. Every one is a constraint you would hit in normal use.

etcd

  • It is sized for coordination data, not application data: the default backend quota is 2 GB and 8 GB is the documented recommended maximum, and exceeding it puts the cluster into a NOSPACE alarm where it accepts no writes until an operator compacts, defragments and clears the alarm by hand.
  • Every write is replicated and fsynced before acknowledgement, so cluster performance is bounded by the slowest disk in it; a member on network-attached storage with high fsync latency causes leader elections and cluster-wide latency spikes that look like network problems and are not.
  • Adding members increases availability but reduces write throughput, because each write must reach a larger quorum; you run three or five members for fault tolerance, and increasing capacity means faster hardware rather than more nodes.
  • There is no sharding and no multi-tenancy, so isolating workloads means running separate clusters, each with its own quorum, certificates, backup schedule and upgrade path, and that operational multiplication is often unexpected.
  • Running it yourself is a real job: periodic compaction and defragmentation, snapshot backups you have actually rehearsed restoring, and rotation of both peer and client TLS certificates, none of which happens automatically outside a managed Kubernetes service.
  • Losing quorum is not self-healing; recovering a cluster that has lost a majority means restoring from a snapshot and accepting that everything written since that snapshot is gone, which makes backup frequency a data-loss budget decision rather than a routine setting.

Thought Machine

  • A core replacement is a two to four year programme at an established bank, and the licence is a minority of total cost against system integrator fees, parallel running and data migration.
  • The smart contract model presumes a bank can write and maintain Python financial products, and institutions without that engineering capability end up outsourcing the very flexibility they bought.
  • Cloud infrastructure cost sits with the bank and is not trivial at scale, so total cost of ownership comparisons against a hosted legacy core often miss a large recurring line.
  • Thought Machine has made two rounds of job cuts pursuing profitability, which is a legitimate supplier stability concern for a system a bank expects to run for fifteen years.
  • The functional footprint is core ledger and product engine, so origination, collections, regulatory reporting and channels all come from other vendors, and the integration estate around Vault is the bank's problem to design and own.

Pricing, plan by plan

etcd

On request

No published plan breakdown. See the etcd review.

Thought Machine

On request
  • Vault Core$undefined/year
    • Quoted per institution, typically on account volumes or annual contract value
    • Implementation and system integrator costs commonly exceed the licence fee
    • Cloud infrastructure costs are the bank's and are not included

Which should you pick?

Choose etcd if

  • You need raft consensus.
  • You also want linearizable reads.
etcd review

Choose Thought Machine if

  • You need smart contract product engine.
  • You work on Web, API, Cloud.
  • You also want cloud native architecture.
Thought Machine review

Questions people ask

Is etcd or Thought Machine better?
Neither clearly leads. etcd starts at On request and Thought Machine at On request, and user ratings are close enough to be indistinguishable. Choose on capability and platform support.
Which is cheaper, etcd or Thought Machine?
etcd starts at On request and Thought Machine at On request.
Does etcd or Thought Machine run on more platforms?
etcd runs on Web. Thought Machine runs on Web, API, Cloud.
What is etcd best used for?
etcd is most often used for storing kubernetes cluster state, which is what the overwhelming majority of etcd deployments are doing, leader election and distributed locking in a home-grown scheduler or control plane, using leases and transactions, service discovery and dynamic configuration where readers need to be notified of changes rather than poll for them, coordinating failover in a clustered database, as patroni does for postgresql. Of those, storing kubernetes cluster state, which is what the overwhelming majority of etcd deployments are doing and leader election and distributed locking in a home-grown scheduler or control plane, using leases and transactions are not what Thought Machine is typically brought in for.
What can etcd do that Thought Machine cannot?
etcd covers Raft consensus, Linearizable reads, Transactions, Leases. Thought Machine covers Smart contract product engine, Cloud native architecture, Real time ledger, Vault Payments.

Answered from the vendors’ own pages

etcd: Can I use etcd as an application database?

No. It is designed for metadata and coordination, with a recommended maximum store size of around 8 GB, no sharding and a write path deliberately optimised for durability rather than throughput. Application data belongs in a database built for it.

Thought Machine: Is Vault Core genuinely cloud native?

Yes. It is containerised, runs on Kubernetes on public cloud, and posts in real time rather than in overnight batch.

etcd: How many members should a cluster have?

Three for most cases, five where you need to survive two simultaneous failures. Always an odd number, because an even-sized cluster gains no additional fault tolerance while making quorum harder to reach.

Thought Machine: How long does a migration take?

Eighteen months at the absolute fastest for a narrow greenfield launch; two to four years for a phased migration at an established bank.

etcd: What happens when the store fills up?

The cluster raises a NOSPACE alarm and stops accepting writes, becoming effectively read-only. Recovery requires compacting old revisions, defragmenting each member and then explicitly disarming the alarm, all done by an operator.

Thought Machine: Does the licence fee represent the total cost?

No. Implementation, system integration, parallel running and cloud infrastructure typically cost more than the licence over the programme.

etcd: Why is my etcd cluster slow or unstable?

Almost always disk latency. Because every write is fsynced before acknowledgement, slow or shared storage causes heartbeat timeouts, leader elections and cascading latency. Local SSDs with low fsync latency are effectively a requirement.

etcd: How does it compare with Consul or ZooKeeper?

All three provide consensus-backed coordination. etcd has the simplest data model and the Kubernetes ecosystem behind it; Consul bundles service discovery, health checking and a service mesh; ZooKeeper is older, JVM-based and still common under Kafka and Hadoop-era systems.

Share

Related pages

More on etcd

More on Thought Machine

Keep looking

Other head to heads