Snapshot of server
GET /snapshot/:hash/server/:vendor
Snapshot of the server table for one vendor. A vendor without rows in the table gets empty arrays.
Snapshots let you read several tables from the same version of the database:
every file of one {hash} comes from the same database, so references between them
always resolve, e.g. every price's server is in the server snapshot of that hash.
Pin database_hash from /healthcheck at the start of a refresh, and fetch every file
with it. /snapshot/{hash} lists the files with checksums, so you can skip the ones
that didn't change since your last refresh.
Contents:
- Only available records: status
active or planned-for-retirement. Inactive and retired records are left out, and so are rows that reference them (e.g. the prices of a retired server, or of an inactive zone).
- Every column of the table except
observed_at: the hash already identifies the data version. For per-row timestamps, use /table/*.
- Sorted by the table's primary key.
- Columnar JSONJSON: JavaScript Object NotationA text format for structured data, built from objects (key-value pairs), arrays, strings, numbers, booleans, and null. Most languages and tools read and write it, which makes it a common format for API responses and benchmark output.JSONGlossary: one array per column, with one element per row, all arrays of the same length.
The response is immutable for a hash, requires an APIAPI: Application Programming InterfaceA defined set of requests and responses through which one program uses another, such as a web service's HTTP endpoints or a library's functions.WikipediaGlossary key, and is sent gzipped only:
send Accept-Encoding: gzip (e.g. curl --compressed), or you get a 406. A 503 of
type /problems/snapshot-preparing means the replica is still downloading or building
the hash: retry the same request after Retry-After seconds (for about a minute at
most). A 404 of type /problems/snapshot-not-available means the hash isn't
available: switch to the current hash in the response body, and fetch again the files
you already fetched. A 404 of type /problems/not-found is an unknown table, vendor
or id: don't retry.
Request
hash (path parameter, required): Database hash, e.g. `database_hash` of `/healthcheck`. vendor (path parameter, required): Vendor id, e.g. `aws`.
Hash not available (problem type /problems/snapshot-not-available), or unknown table, vendor or malformed id (/problems/not-found).
The request doesn't accept gzip.
The hash is being prepared on this replica (problem type /problems/snapshot-preparing): retry the same request after Retry-After seconds.
200: Successful Response. vendor_id: Reference to the Vendor. server_id: Unique identifier, as called at the Vendor. name: Human-friendly name. api_reference: How this resource is referenced in the vendor API calls. This is usually either the id or name of the resource, depending on the vendor and actual API endpoint. api_reference_object: How this resource is referenced in the vendor API calls, including the parameter name(s). display_name: Human-friendly reference (usually the id or name) of the resource. description: Short description. family: Broad workload class of the server in the vendor's own naming, e.g. General Purpose, Compute Optimized or Memory Optimized. series: Hardware line and generation of the server, e.g. m5 (AWS), n2 (GCP) or b3 (OVH). vcpus: Default number of virtual CPUs (vCPU) of the server. hypervisor: Hypervisor of the virtual server, e.g. Xen, KVM, Nitro or Dedicated. cpu_allocation: Allocation of CPU(s) to the server, e.g. shared, burstable or dedicated. Values: Shared, Burstable, Dedicated. cpu_cores: Default number of CPU cores of the server. Equals to vCPUs when HyperThreading is disabled. cpu_speed: Vendor-reported maximum CPU clock speed (GHz). cpu_architecture: CPU architecture (arm64, arm64_mac, i386, or x86_64). Values: arm64, arm64_mac, i386, x86_64, x86_64_mac. cpu_manufacturer: The manufacturer of the primary processor, e.g. Intel or AMD. cpu_family: The product line/family of the primary processor, e.g. Xeon, Core i7, Ryzen 9. cpu_model: The model number of the primary processor, e.g. 9750H. cpu_l1d_cache: L1 data cache size (KiB). cpu_l1d_cache_total: Total L1 data cache size (KiB) across all cores. cpu_l1i_cache: L1 instruction cache size (KiB). cpu_l1i_cache_total: Total L1 instruction cache size (KiB) across all cores. cpu_l2_cache: L2 cache size (KiB). cpu_l2_cache_total: Total L2 cache size (KiB) across all cores. cpu_l3_cache: L3 cache size (KiB). cpu_l3_cache_total: Total L3 cache size (KiB) across all cores. cpu_flags: CPU features/flags. cpus: JSON array of known CPU details, e.g. the manufacturer, family, model; L1/L2/L3 cache size; microcode version; feature flags; bugs etc. manufacturer: Manufacturer. family: Family. model: Model. cores: Cores. threads: Threads. l1_cache_size: L1 Cache Size. l2_cache_size: L2 Cache Size. l3_cache_size: L3 Cache Size. microcode: Microcode. capabilities: Capabilities. bugs: Bugs. bogomips: Bogomips. ecpus: The effective "real-world" core count, calculated by dividing the maximum multi-core SCore by the single-core SCore. scalability: Measures how efficiently the server scales from a single core performance to using multiple cores. A score of 100% means perfect linear scaling with zero performance loss. hw_virt: If hardware virtualization (e.g. KVM) is supported. memory_amount: RAM amount (MiB) reported by the vendor. memory_amount_actual: Actual RAM amount (MiB) measured on the instance via lstopo or other tool. This amount might not match the vendor-reported memory due to the BIOS or the hypervisor reserving a small percentage. memory_generation: Generation of the DDR SDRAM, e.g. DDR4 or DDR5. Values: DDR3, DDR4, DDR5. memory_speed: DDR SDRAM clock rate (Mhz). memory_ecc: If the DDR SDRAM uses error correction code to detect and correct n-bit data corruption. accelerator_count: Number of accelerator(s), e.g. GPUs or NPUs. accelerator_memory_min: Memory (MiB) allocated to the lowest-end accelerator. accelerator_memory_total: Overall memory (MiB) allocated to all the accelerator(s). accelerator_type: The type of the primary accelerator, e.g. GPU or NPU. Values: GPU, TPU, NPU, FPGA, VPU. accelerator_manufacturer: The manufacturer of the primary accelerator, e.g. Nvidia or AMD. accelerator_family: The product family of the primary accelerator, e.g. Turing. accelerator_model: The model number of the primary accelerator, e.g. Tesla T4. accelerators: JSON array of accelerator details, including the manufacturer, name, and memory (MiB) of each accelerator. memory: Memory. firmware_version: Firmware Version. bios_version: Bios Version. graphics_clock: Graphics Clock. sm_clock: Sm Clock. mem_clock: Mem Clock. video_clock: Video Clock. storage_size: Overall size (GB) of the disk(s). storage_type: Primary disk type, e.g. HDD, SSD, or NVMe SSD. Values: hdd, ssd, nvme ssd. storages: JSON array of disks attached to the server, including the size (GB) and type of each disk. size: Size. storage_type: Type of a storage, e.g. HDD or SSD. Values: hdd, ssd, nvme ssd. description: Description. compatible_storage_ids: List of storage_ids that can be attached to the server as extra storage. network_speed_baseline: The baseline network performance (Gbps) of the server, aggregated over all network cards. network_speed_max: The maximum network performance (Gbps) of the server, aggregated over all network cards. network_storage_speed_baseline: The baseline bandwidth performance of network-attached storage (Gbps). network_storage_speed_max: The maximum bandwidth performance of network-attached storage (Gbps). inbound_traffic: Amount of complimentary inbound traffic (GB) per month. outbound_traffic: Amount of complimentary outbound traffic (GB) per month. ipv4: Number of complimentary IPv4 address(es). average_time_to_start: Average time to start the server (seconds). status: Status of the resource (active or inactive). Values: active, inactive, planned-for-retirement, retired. 401: No API key. 404: Hash not available (problem type `/problems/snapshot-not-available`), or unknown table, vendor or malformed id (`/problems/not-found`). 406: The request doesn't accept gzip. 422: Validation Error. detail: Detail. loc: Location. msg: Message. type: Error Type. input: Input. ctx: Context. 503: The hash is being prepared on this replica (problem type `/problems/snapshot-preparing`): retry the same request after `Retry-After` seconds.