github.com/outbrain/consul@v1.4.5/website/source/intro/vs/skydns.html.md (about) 1 --- 2 layout: "intro" 3 page_title: "Consul vs. SkyDNS" 4 sidebar_current: "vs-other-skydns" 5 description: |- 6 SkyDNS is a tool designed to provide service discovery. It uses multiple central servers that are strongly-consistent and fault-tolerant. Nodes register services using an HTTP API, and queries can be made over HTTP or DNS to perform discovery. 7 --- 8 9 # Consul vs. SkyDNS 10 11 SkyDNS is a tool designed to provide service discovery. 12 It uses multiple central servers that are strongly-consistent and 13 fault-tolerant. Nodes register services using an HTTP API, and 14 queries can be made over HTTP or DNS to perform discovery. 15 16 Consul is very similar but provides a superset of features. Consul 17 also relies on multiple central servers to provide strong consistency 18 and fault tolerance. Nodes can use an HTTP API or use an agent to 19 register services, and queries are made over HTTP or DNS. 20 21 However, the systems differ in many ways. Consul provides a much richer 22 health checking framework with support for arbitrary checks and 23 a highly scalable failure detection scheme. SkyDNS relies on naive 24 heartbeating and TTLs, an approach which has known scalability issues. 25 Additionally, the heartbeat only provides a limited liveness check 26 versus the rich health checks that Consul performs. 27 28 Multiple datacenters can be supported by using "regions" in SkyDNS; 29 however, the data is managed and queried from a single cluster. If servers 30 are split between datacenters, the replication protocol will suffer from 31 very long commit times. If all the SkyDNS servers are in a central datacenter, 32 then connectivity issues can cause entire datacenters to lose availability. 33 Additionally, even without a connectivity issue, query performance will 34 suffer as requests must always be performed in a remote datacenter. 35 36 Consul supports multiple datacenters out of the box, and it purposely 37 scopes the managed data to be per-datacenter. This means each datacenter 38 runs an independent cluster of servers. Requests are forwarded to remote 39 datacenters if necessary; requests for services within a datacenter 40 never go over the WAN, and connectivity issues between datacenters do not 41 affect availability within a datacenter. Additionally, the unavailability 42 of one datacenter does not affect the discovery of services 43 in any other datacenter.