A matter takes too long to open. A document saves slowly. A remote desktop pauses while an attorney is trying to finish a filing. Your systems are available, but the experience is still interrupting work. Moving an application to the cloud does not automatically make it faster. Its performance depends on the entire path between the user, the application and the data it needs. An internet connection, a distant hosting region, an overloaded server or a slow database can each create delays. The most useful first step is to identify where the waiting happens. That gives your firm a way to fix the cause before paying for a larger server or another migration.
Availability and performance measure different things
An application can be online and still respond too slowly for productive work. Uptime measures availability. Performance measures how quickly users can complete tasks.
Define a few tasks that matter to your firm: opening a matter, searching documents, saving a file and running a billing report. Record how long each takes from different offices and remote locations. Use those measurements alongside infrastructure monitoring so the investigation reflects what attorneys and staff actually experience.
A fast internet plan does not guarantee a fast application
Bandwidth describes how much data a connection can carry. Latency is the delay involved in communication; round-trip time measures a request traveling out and back. Packet loss and variation in delay can also affect the experience.
A large download may benefit from more bandwidth. An interactive application that makes many sequential requests can be sensitive to latency even when it sends relatively little data. Microsoft identifies latency, packet loss and network conditions as important considerations when investigating Azure Virtual Desktop connection quality. [1]
For a fictional illustration, imagine a workflow that needs 40 sequential network round trips. At 70 milliseconds per trip, its network waiting component is 2.8 seconds. At 20 milliseconds, that component is 0.8 seconds. The two-second difference is arithmetic, not a prediction for a particular application; processing time, parallel requests and caching change real results.
Test the route to the affected service during busy periods. A speed test to a nearby test server does not establish that the path to your cloud application is healthy.
The connection may be taking an unnecessary detour
Remote users sometimes connect to an office VPN before reaching an application in the cloud. That route can add distance or make the office firewall and internet circuit a shared bottleneck.
Compare how traffic travels from each office and from remote devices. Check whether cloud-bound traffic is routed through a central office, whether the VPN device reaches its processing limits and whether transfers compete with interactive sessions.
Microsoft's performance guidance includes optimizing network paths and reducing unnecessary communication. [2] A shorter route can help, but changes should preserve the access controls and inspection your firm requires. Test approved alternatives rather than removing protection to make a speed test look better.
Hybrid systems can create repeated trips between environments
A cloud-hosted application may still depend on an office-based database, file share or identity service. Those dependencies can turn a simple task into repeated journeys across the connection between the two environments.
For example, moving an application server while leaving its frequently accessed database in the office can expose a previously local interaction to network delay. A faster cloud server cannot remove that delay on its own.
Hybrid cloud needs connectivity appropriate to those dependencies. Site-to-site VPN and private connections such as Azure ExpressRoute are options; hybrid cloud is not limited to VPN. [3]
Before upgrading the link, consider whether tightly coupled application and database components should be hosted together. Where the software vendor supports it, placing the application session near its data may be a better design than making every user interaction cross the office-to-cloud connection.
Your hosting region may be too far from the work
Choosing a region closer to users can reduce network latency. Microsoft recommends evaluating user locations, service availability and regional requirements when selecting Azure regions. [4]
For a California-based firm, nearby western US regions are reasonable candidates to test. The best choice still depends on measured routes, where staff work and where the application's data resides.
Avoid evaluating user proximity alone. Moving an application closer to attorneys while leaving its database far away may exchange one delay for another. Evaluate the user-to-application and application-to-data paths together. Microsoft's published regional latency statistics describe communication between Azure regions, not the complete route from your offices. [5]
Compute and storage may be limiting the workload
Resource sizing should reflect busy periods, concurrent users and background processing. Review CPU demand, memory pressure, queues and task duration around the time a slowdown occurs. Infrastructure optimization includes selecting resources that fit the workload rather than assuming any cloud instance will perform equally. [2]
Storage deserves a separate check. Capacity in gigabytes tells you how much a disk can hold; it does not tell you how quickly it serves requests. In Azure, virtual machines and attached disks have their own input/output and throughput limits. A faster disk may still be constrained by the VM's limit. [6]
Review both layers before upgrading. Check whether background indexing, imports or backup activity overlaps with busy application periods. An increase in capacity should address an observed constraint.
The application or database may need tuning
Not every delay originates in the cloud network. Microsoft documents database bottlenecks including missing indexes, inefficient query plans, resource waits and blocking. [7]
If one report is slow while document access works normally, investigate that report and its database activity. If every action slows down for all users, examine shared dependencies. These are investigation clues, not diagnoses.
Work with the application vendor before changing database behavior, caching or deployment architecture. A technically possible configuration is not necessarily one the vendor supports.
Use the symptoms to focus the investigation
| What users experience | Possible cause | What to investigate |
|---|---|---|
| One office is slow while others work normally | Local network, firewall or internet path | Compare wired connections, utilization and the route to the service |
| Remote users are slower than office users | Home connectivity or an indirect VPN route | Compare paths, packet loss and VPN device demand |
| All users slow down during busy periods | Shared compute, storage or connection limits | Correlate task delays with resource and traffic metrics |
| Documents open slowly but other tasks are responsive | File-service or storage bottleneck | Measure file operations and inspect storage latency and limits |
| A particular search or report is slow | Application or database issue | Review query duration, waits and vendor diagnostics |
| Delays began after a partial migration | New cross-environment dependency | Map application, database, identity and file-service traffic |
| Sessions pause intermittently | Variable delay, packet loss or resource contention | Compare timestamps across network and host monitoring |
Fix the measured bottleneck and retest
Start with a repeatable baseline from affected users. Correlate slow tasks with network and infrastructure measurements. Then make a controlled change and run the same tasks again, including during normal busy periods.
A suitable remedy may be a better route, adjusted storage, application tuning, a different region or a redesigned hybrid dependency. Bigger servers and faster circuits can help when those are the limiting factors. They should follow the diagnosis.
Allnext Legal helps law firms assess cloud connectivity, application hosting and day-to-day performance so technology supports the work your people need to complete.
Sources
- Microsoft — Troubleshoot Azure Virtual Desktop connection quality
- Microsoft — Optimize code and infrastructure for performance
- Microsoft — Hybrid connectivity options
- Microsoft — Select Azure regions
- Microsoft — Azure network round-trip latency statistics
- Microsoft — Virtual machine and disk performance
- Microsoft — Identify database query performance bottlenecks
Resolve Your Cloud Performance Issues.
Contact us
