Load and performance testing
Driving traffic at infrastructure you own, from a controlled egress, at a volume no consumer connection could produce. The target is yours, so the registry entry is nobody's business but your own.
Addresses that belong to a hosting company, on hardware built for throughput. They are fast and cost little to run per address, and any site that looks up the registration behind an IP can see it belongs to a hosting network. We do not sell them yet.
This page explains the product category. ProxyJam's catalogue sells static addresses registered to named internet providers, which is a different thing on the one axis that decides most jobs.
In short: A datacenter proxy is an address held by a hosting company: fast, built for volume, and publicly registered as exactly what it is. That makes it the right tool where nobody checks the network — your own systems, APIs that authenticate you, allow-listed egress — and a poor one for consumer-facing sites that screen out hosting ranges.
A datacenter proxy is an IP address that belongs to a hosting company rather than to a consumer internet provider. It lives in a data centre, it answers at server speed, and the public registry says what it is.
That last part decides most jobs. Any site that cares who is connecting can look up the holder of an address, and some sites refuse hosting ranges or serve them different content as a matter of policy. For work where nobody is checking, that costs you nothing: internal testing, your own infrastructure, public data served to anyone who asks. For work where a refused or altered response looks just like a real one, it can cost you the whole run.
IP addresses are handed down a registry chain. IANA allocates blocks to the Regional Internet Registries, and they serve the networks and organisations below them (RFC 7020). Each registry publishes who holds what. The modern way to read that record is RDAP, a JSON-over-HTTPS alternative to the older text-based WHOIS, and ARIN’s RDAP documentation describes what a query for an address returns: the network’s name, its registration type and the entities attached to it, including the registrant. A client does not have to know which registry to ask, because IANA publishes bootstrap files that map address blocks to the right registry’s service (RFC 9224).
Commercial IP-intelligence services build on that record and label ranges by use — hosting, business, consumer broadband, mobile. Each keeps its own data and method, so we do not name one; what they share is the public registration underneath.
Nothing here is hidden or subtle. The economics and the weakness both follow from one fact about how the address is registered.
A hosting company holds a block of addresses. It obtains the block through the regional registry system, routes it from its own network, and the registration record naming the holder is public and free for anyone to query.
Those addresses cost little to operate at scale. One rack can answer for a great many of them, the transit is already paid for, and nothing depends on a household router staying plugged in. That is why a datacenter address usually costs less per address than one sourced from a consumer network.
Any target can do the lookup you can. A site that cares who is asking reads the registration behind the address and gets a hosting company’s name back. Nothing about the request is disguised — the honesty of the record is the product’s defining property, useful in some jobs and fatal in others.
| Shared datacenter | Dedicated datacenter | Static ISP | |
|---|---|---|---|
| Who else uses it | Several customers at once | You alone | You alone |
| Relative cost per address | Lowest of the three | Low | Highest of the three |
| Throughput ceiling | Shared, so variable | The port's, effectively | The port's, effectively |
| Reputation you inherit | Whatever your neighbours did | Your own, from day one | Your own, from day one |
| What a network lookup returns | A hosting company | A hosting company | A named internet provider |
| Where it stops working | Sites that screen out hosting networks | Sites that screen out hosting networks | Not on this axis — a lookup returns an internet provider |
Driving traffic at infrastructure you own, from a controlled egress, at a volume no consumer connection could produce. The target is yours, so the registry entry is nobody's business but your own.
When access is decided by a key or a signed request, the network behind the address is irrelevant. Paying residential prices to reach an endpoint that already knows who you are is money spent on nothing.
A partner or an internal service that admits a named address does not care where that address lives. It cares that the address does not change, and a dedicated datacenter address is a simple way to promise that.
Mirroring, backups, asset fetching and anything else that moves a great deal of data between systems you control. Bandwidth here is a line item, and server-grade transit is built to carry it.
Checking that your own endpoints answer from outside your network. What you need is a vantage point and a stable one, not a disguise.
Routing a build agent or a test suite through a known address so firewall rules can be written once. Predictable, and no part of the job depends on looking like a person.
Server-grade hardware on paid-for transit, with no household router or radio link in the path. For work that does not care who is asking, that is most of the decision.
The block is registered to a hosting company and announced from its network, and both facts are public. A target that treats hosting ranges as automated traffic is not guessing — it is reading a record you cannot edit.
An address hosted on server infrastructure but registered to an internet provider keeps the server side of the arrangement and changes what the lookup returns. That is the product ProxyJam actually sells.
Whether a datacenter address is enough for a build agent, a test suite or an API that checks a key, or whether the job needs an address registered to an internet provider.
Whether a datacenter address is enough for load tests, allow-listed egress or monitoring your own endpoints, or whether the job needs an address registered to an internet provider.
Open What Is My IP in a browser that uses the connection, and it shows the address with its internet provider and network (ASN).
Open https://rdap.arin.net/registry/ip/ADDRESS in a browser, with the address in place of ADDRESS. If another regional registry holds the block, ARIN redirects the query there.
Read the name field and the entity whose role is registrant. If that is a hosting company, or a customer of one, that is the same public record any site can read.
An endpoint you own, an API that checks a key, or a partner that allow-lists your address does not; a consumer-facing site that varies what it shows by network may.
If it does, start from an address registered to an internet provider, not from more careful requests on a hosting address.
This is a category explainer, not an offer. Where an answer states a term of sale it describes the static ISP product, which is what ProxyJam sells, and links to the document that binds us.
It is an IP address that belongs to a hosting company rather than to a home or mobile subscriber, served from that company's own network and hardware. You rent the address for a period, it does not move, and requests leaving through it reach the target at server speed. Everything good and everything limiting about the category follows from the ownership: the record naming the hosting company is public and cannot be changed by the person renting the address.
Because nothing in the supply chain is scarce or awkward. The provider already holds the addresses, already runs the racks and already pays for the transit, so adding one more address to a block costs it little. Consumer addresses have to come from subscriber lines one at a time, and that difference in supply is what shows up on the invoice.
A website that checks will know the address belongs to a hosting network, and on many consumer-facing sites that leads to the same treatment. The check is simple: read the public registration record for the address, which names the organisation holding the block, or consult one of the commercial databases that label address ranges by use. None of this needs clever detection, which is why a claim that a datacenter range is undetectable deserves scepticism.
Whether anyone else is using the address at the same time. Shared costs less and you inherit whatever reputation your neighbours have already built on that address, which you cannot see and cannot undo. Dedicated costs more and the only history on the address is yours. For anything where a block would cost you real time, the difference is not a luxury.
When the target does not care who is asking, or already knows. Testing your own systems, reaching an API that authenticates with a key, moving data between services you control, holding a fixed egress address for a firewall rule. In all of those the network registration is irrelevant, and paying for an internet provider's registration buys you nothing.
Anywhere a consumer-facing site decides what to show based on who appears to be asking. Retail, streaming, social platforms, ticketing, advertising — many of them treat hosting ranges differently from household ones, and no amount of care in how you send the request changes what the registry says about the address it came from.
An address hosted on the same class of server infrastructure but registered to a named internet provider, so a lookup returns that provider rather than a hosting company. It is priced per IP per period rather than per gigabyte, held for the whole period, replaced free up to 20 % of the order per calendar month, and the first order carries a 72-hour refund covering up to 10 IPs that using the proxies does not void. ISP Proxies