Where Is Your HTTPS Traffic Decrypted? A Data Residency Guide for SWG Buyers

Where Is Your HTTPS Traffic Decrypted? A Data Residency Guide for SWG Buyers

Last updated: November 2026

Direct answer: Data residency is the requirement that certain data be stored and processed in a specific country or region. For a secure web gateway (SWG), the practical residency question is where HTTPS traffic is decrypted. A cloud-proxy SWG decrypts it in a vendor point of presence (PoP). An on-device SWG decrypts it on the employee's laptop, then sends the request straight to its destination.

That one distinction decides whether the plaintext of your users' web sessions sits in someone else's data center, and in which country. This guide defines data residency, data sovereignty, and digital sovereignty. It then explains the SWG question most residency guides skip and gives you a list of questions to ask any vendor, us included. We make dope.security, an on-device SWG, so read the product sections with that in mind. Where on-device inspection doesn't settle a residency question, we say so.

Key takeaways

  • Data residency is about where data is stored and processed. Data sovereignty is about whose laws govern it. Digital sovereignty is about who controls the technology it runs on.
  • Turning on SSL inspection (also called TLS inspection or SSL decryption) creates new plaintext. Where that plaintext exists is a residency question, and most residency guides never ask it.
  • A cloud SWG decrypts in a vendor PoP, and failover can move that to another PoP. An on-device SWG decrypts on the managed endpoint.
  • "On-device" does not mean "nothing leaves the device." Logs, AI DLP classification, and SaaS scanning still involve the cloud. Ask every vendor for the full data flow.

What is data residency?

Data residency is the physical or geographic location where an organization's data is stored and processed, usually stated as a country or region. IBM puts it as the location of the data centers, servers, or other infrastructure that store and process the data.

Data residency requirements come from three places:

  1. Law and regulation. Privacy and sector rules can restrict where personal data goes. The EU's GDPR, for example, puts conditions on transferring personal data outside the European Economic Area.
  2. Contracts. Public-sector, healthcare, and financial customers often write residency into their contracts and RFPs ("customer data must remain in the EU").
  3. Internal policy. Many security and privacy teams choose to keep certain data in-region even when no law requires it.

This is not legal advice. Which rules apply to you depends on your data, your customers, and where you operate, so bring your privacy counsel in early. For the PCI DSS and GDPR side of this, see how to meet PCI DSS and GDPR with cloud security.

Data residency vs data sovereignty vs digital sovereignty

The three terms get used interchangeably. They shouldn't be.

Term Plain definition The question it answers What it looks like in an SWG evaluation
Data residency The country or region where data is physically stored and processed. Where does this data live and get processed? "Which region stores our web transaction logs? Where is HTTPS decrypted?"
Data sovereignty The principle that data is subject to the laws of the country where it is collected or stored. Whose laws apply, and who can demand access? "Which jurisdictions can compel the vendor to hand over our logs? Where is support staff located?"
Digital sovereignty A country's or organization's ability to control its own digital infrastructure, data, and technology choices. How dependent are we on providers we don't control? "If this vendor's cloud goes down or changes terms, can we still enforce policy?"

Put simply, data residency is a geographic idea, data sovereignty is a legal one, and digital sovereignty is a strategic one. IBM draws the same line between the first two: residency is geographic, sovereignty is legal, and residency often decides which country's sovereignty rules apply. Digital sovereignty is a broader policy debate, mostly about governments and critical infrastructure (Wikipedia overview). It comes up in SWG deals as a question about dependence on a single foreign cloud.

One more term you'll hear on vendor calls is sovereign SASE. It generally means a cloud security vendor offering to keep inspection inside a given jurisdiction, usually through in-country PoPs. That reduces where your plaintext can go. It doesn't change the fact that a third party decrypts it. Inspection on the device works differently: decryption never moves to a vendor facility in the first place.

The SWG question most residency guides skip: where is HTTPS decrypted?

Most web traffic is encrypted, so an SWG can't inspect it without breaking TLS open. SSL inspection decrypts the session, inspects the plaintext, and re-encrypts it before it goes on. Our SSL inspection explainer walks through the mechanics. For residency, the question is simple: where does that plaintext exist, even briefly?

There are only two places to do it.

Architecture 1: decrypted in a vendor point of presence

A cloud-proxy SWG steers device traffic to the vendor's nearest PoP. The PoP terminates TLS, inspects the request (URL, file upload, AI prompt), applies policy, and forwards it to the destination. The plaintext of every inspected session exists in that vendor's data center.

What that means for residency:

  • The PoP's country matters. If your nearest PoP is across a border, your users' decrypted sessions cross it too.
  • Failover moves the plaintext. When a PoP is overloaded or down, traffic typically shifts to the next available one, which may be in another country. You have to account for where decryption happens during an incident as well as on a normal day.
  • The vendor becomes a data processor for decrypted content. Under GDPR that is a processing relationship to document, along with the vendor's subprocessors.
  • Some regions need special infrastructure. Mainland China is the clearest example. Cloud-proxy vendors generally need separate in-country arrangements to work there, and many sell them as add-ons. See which SWG works in China without a paid add-on and why Zscaler struggles at China sites.

None of this makes a cloud SWG non-compliant. Plenty of organizations meet residency requirements with in-region PoPs and a strong DPA. It does mean the inspection plane is part of your data map, and you have to keep checking it.

Architecture 2: decrypted on the device

An on-device SWG runs the proxy on the endpoint itself. TLS is decrypted, inspected, and re-encrypted on the managed laptop, and the request then goes directly to its destination. No vendor PoP sits in the inspection path for that traffic.

What that means for residency:

  • The plaintext stays where it already was. The laptop had the plaintext anyway, because that's where the user typed it. Inspecting there adds no new party and no new country.
  • There's no inspection failover to track. No PoP means no PoP failover. If the employee is in Frankfurt, the decryption happens in Frankfurt.
  • It works in restricted geographies. There is no vendor PoP to reach, so the same agent works on hotel Wi-Fi, in a branch office, or in a country where a cloud proxy struggles.

We've made this architecture argument before in Zero Trust Shouldn't Mean Routing Everything Through a Third Party and in our comparison of endpoint-based vs cloud-proxy SWGs. On-device inspection settles the decryption question. It doesn't settle every residency question, which is why the next section matters.

What still goes to the cloud with an on-device SWG?

"On-device" is not a synonym for "nothing ever leaves." Here's what goes where with dope.security today, based on our public DPA cover letter, sub-processor list, and product documentation.

Data Where it is processed What leaves the device
Decrypted web traffic (page content, uploads, prompts) On the endpoint, inside dope.endpoint None of it goes to dope.security for decryption. Customer web traffic does not transit our cloud.
Transaction records (user, URL, timestamp, size, policy result) dope.console analytics Uploaded to the region the customer chooses (the DPA lists more than five options). Deleted automatically within 30 days at most. Can be synced to an Amazon S3 bucket you own for SIEM or longer retention.
User and policy data (device name, username/email, groups, IP addresses) dope.console Stored in an active-active, multi-region database for failover and performance.
Dopamine DLP classification (file uploads and AI prompts) Text extracted on the device, classified in the dope.cloud The extracted text needed for classification is sent to the dope.cloud, which uses OpenAI's zero-data-retention API. The original file stays on the device. Content is not retained or used for training. A verdict (Sensitive / Not Sensitive) comes back and the endpoint enforces it.
DLP detections dope.console Each Block or Monitor event is logged with a short, plain-language Dopamine Summary of why it was flagged.
Anti-malware Endpoint computes file hashes, then checks a malware service File hashes are checked against a malware service. Verdicts show up in dope.console.
CASB Neural (Microsoft 365 and Google Drive) In the cloud, via API This is data at rest in your SaaS tenant, a separate plane from web traffic. Files that need review are downloaded, classified by an LLM, and deleted. File content is not stored.

Three honest notes:

  • Dopamine DLP is the one place web content leaves the endpoint. It's a minimal extract, under zero retention, but it does leave. If prompts or uploads to AI tools fall under your residency rules, ask us where classification runs and include it in your review.
  • Our subprocessors include OpenAI and AWS. Both are listed on our sub-processor list as hosted globally. Check them against your requirements as you would for any vendor.
  • Sovereignty questions remain. dope.security is a US company, and our DPA cover letter says support is provided from the US, the EU, and India. On-device inspection doesn't change who can be asked for logs or where support staff sit. That's a contract question.

On the compliance side, dope.security is SOC 2 Type 2 certified, and our DPA includes contractual clauses for EU personal data transfers. dope.security does not replace your SaaS providers' region settings, your cloud residency map, or legal advice.

What to ask any SWG vendor, including dope.security

Ask for a data-flow diagram, then work through this list. A vendor confident in its architecture will answer every line without hesitation.

  1. Where is TLS decrypted? In which facility and which country, under normal operation?
  2. Where is it decrypted during failover? Name the backup PoPs or regions, and say whether you can pin or exclude them.
  3. What metadata and logs leave the device or the PoP? List the fields: user, URL, IP, file names, and whether page content is ever included.
  4. Where are logs stored, and for how long? Is the region customer-selectable? Can you export to storage you own?
  5. Where does DLP classification run? On the device, in the vendor cloud, or with a third-party model? Under what retention and training terms?
  6. Who are the subprocessors, and where are they hosted?
  7. Where are support and engineering staff who can access customer data located?
  8. What happens to enforcement if the vendor cloud is unreachable?

We gave buyers the same advice in our on-device privacy post: trust, then verify the data flow, and apply that to every vendor, including this one. For more on evaluating SWGs in regulated environments, see the best secure web gateway for healthcare, finance, and regulated teams.

When on-device inspection helps, and when it doesn't

It helps when:

  • Your fleet is managed Mac and Windows laptops, including remote staff and frequent travelers.
  • You have employees in restricted geographies where reaching a cloud PoP is slow, unreliable, or sold as an add-on.
  • You want to keep a third-party inspection plane out of your data map entirely.
  • Your auditors want a short, checkable statement: decrypted web traffic is inspected on the managed endpoint and doesn't transit the vendor's cloud.

It doesn't, by itself, cover:

  • Data at rest in SaaS. Files already shared out of Microsoft 365 or Google Drive need tenant scanning, which is what CASB Neural does through the cloud.
  • Unmanaged, BYOD, mobile, IoT, or guest devices. dope.security runs on managed Mac and Windows endpoints. Anything that can't run the agent needs a complementary control.
  • Private-app access. dope.security has no built-in ZTNA today. If you need to broker access to internal apps, that's a separate tool.
  • Your vendors' own residency. Where Salesforce or your cloud provider stores data is set in those contracts, not by your SWG.

If the on-device model fits, the dope.SWG product page covers the architecture in more detail, or you can book a 20-minute demo and walk through the data flow with our team.

Frequently asked questions

What is the difference between data residency and data sovereignty?

Data residency is where data is physically stored and processed. Data sovereignty is which country's laws govern that data and who can lawfully demand access to it. Residency is geographic, sovereignty is legal, and the residency location often decides which sovereignty rules apply.

What is digital sovereignty?

Digital sovereignty is a country's or organization's ability to control its own digital infrastructure, data, and technology choices without depending on providers outside its control. It's broader than data sovereignty and is usually discussed at the level of national policy and critical infrastructure.

Does SSL inspection break data residency?

It can. SSL inspection creates a decrypted copy of the traffic wherever the inspection happens. If that's a vendor PoP in another country, decrypted sessions cross borders, including during failover. If inspection happens on the user's device, the plaintext stays where the user is.

Can a cloud SWG keep inspection in one country?

Sometimes. Some vendors offer in-region or "sovereign" PoPs that keep decryption inside a jurisdiction. Confirm the PoP locations, the failover behavior, and whether you can pin traffic to them in writing. Even then, a third party still decrypts your traffic. It just does so in-country.

Does on-device inspection mean no data leaves the laptop?

No. With dope.security, decryption and inspection happen on the endpoint and web traffic goes direct to its destination. Transaction logs still go to dope.console in a region you choose. Dopamine DLP sends extracted text to the dope.cloud for zero-retention classification. CASB Neural scans Microsoft 365 and Google tenants in the cloud. Ask any vendor for the full data flow.

What are typical data residency requirements for a security vendor?

Common requirements include a customer-selectable storage region for logs, a defined retention period, a published subprocessor list, a DPA with transfer clauses, and a clear statement of where decrypted content is processed. Your specific requirements depend on your regulators and contracts. This isn't legal advice.

Related reading

Questions about where your data goes with dope.security? Email sales@dope.security and ask for the data-flow walkthrough.

Compliance
Compliance
Secure Web Gateway
Secure Web Gateway
Data Loss Prevention
Data Loss Prevention
← back to blog Home