Cloud performance tests reveal the impact of location
Performance is one of the most important considerations for healthcare IT professionals when considering the feasibility moving their EHR system to the cloud.
If a cloud-based EHR isn’t fast enough to meet the needs of hospitals and other medical facilities, quality of care will suffer, clinicians and support staff will become frustrated, and provider organizations operating on thin margins will have wasted valuable IT dollars on inadequate technology.
Several industry trends make it clear that providers are increasingly confident cloud-based EHRs can deliver data and applications at sufficient speeds because vendors have dramatically reduced latency. Epic Systems, the largest EHR vendor in the U.S., began offering cloud services in 2014 to medical groups and small hospitals. The company said in 2016 that it has seen a big shift among its customers toward cloud-based systems. One of those customers is University of California San Diego Health, which announced in August 2017 that it is migrating its Epic EHR to the cloud.
Cerner has also revealed that a number of healthcare systems are moving onto its hosted cloud environment. Inspira Health Network, a nonprofit healthcare organization that serves communities across New Jersey, and Bay State, an integrated health system in western Massachusetts health system, have both moved to Cerner’s cloud hosting model.
Another EHR vendor, Athenahealth, offers only cloud-based EHR services to its network of 56 hospitals and 106,000 providers that serves more than 100 million patients. Ellenville Regional Hospital in upstate New York, reports that running its EHR on a single network gives staff in all departments real-time access to patient records.
For health systems that will select their own cloud host, after considering the cost of strategic planning and security, the success of a migration ultimately rests on performance. Can EHR data, applications and services be delivered fast enough to support the needs of clinicians? Cloud performance is generally measured by average latency which represents the delay between the time when a client computer requests data and the cloud platform responds.
Tests have shown that distance between the cloud provider and the enterprise can have a significant impact on latency with delays of as much as 50 percent when the cloud is at a great distance.
Cedexis tested services and found distance was a leading indicator of performance
Monitoring company Cedexis performed tests throughout the United States that quantify this. Tests were performed on all of the major cloud platform vendors. The determining factor in latency performance was distance from the test client system to the servers at the cloud data center, with a deterioration of as much as 50 percent seen over longer distances within the same region.
For example, tests performed in the northwest region recorded latency was as low as 63 ms. when the cloud was near; a latency of 92 ms. was recorded on a cloud system at a greater distance from the test location. A test on cloud platforms in the northeast found 66 ms. latency on the low end and 78 ms. on the high end.
A number of vendors offer cloud performance tests that are free to be used during an evaluation process. One example is CloudHarmony who offers a free service that will test performance for many of the leading vendors at a variety of their geographic locations. The test, at Cloud Harmony Speed Test will provide results for DNS query, downlink and latency.
Migrating an EHR system to the cloud rarely is an all-or-nothing process. Rather, many healthcare IT managers are moving incrementally, evaluating the success of each application migration, learning from mistakes, and carefully preparing the next move based on need and experience. For example, some may decide to first migrate back-office or HR applications, and eventually, migrate clinical apps and billing data at a later date. Providers certainly can choose a one-shot migration to the cloud, but they run the risk of network and system disruptions, loss of productivity, and bandwidth limitations.
After cloud migration goals are set, computing models are chosen, and a migration plan established, healthcare providers can begin choosing a CSP (or more than one). For providers with concerns about potential latency issues, selecting a CSP with a hosted location close to the provider’s facilities makes sense. The provider’s IT staff and the CSP can then begin the cloud migration. This process involves addressing several integration challenges.
Perhaps the primary challenge is to prevent disruptions to systems or services during or as a result of the migration that would impact clinical care, staff productivity, or IT processes. Other challenges include protecting and backing up migrated data and connecting to and integrating disparate systems.
Integration may involve linking cloud-based apps and data with non-cloud apps in legacy systems. “An illustrative scenario could involve a multi-hospital operation which chooses to retain on-premises EHR for inpatient operations but wants to leverage public cloud services for geographically distributed outpatient clinics,” according to the Cloud Standards Customer Council (CSCC).
The CSCC argues that a successful cloud EHR migration depends on security and network connectivity.
“Whether you are ensuring insurance coverage for the public, developing the next generation of cancer drugs, or providing critical care/tier I trauma services, the new emphasis is being put on providing network availability, performance and security,” CSCC writes. “Although creating a highly available network might be expensive, those costs can be offset by the capabilities provided to the organization.”
The vast majority of healthcare providers today have moved or are moving applications, infrastructure, or development platforms to the cloud because they recognize the performance benefits and cost savings. For a cloud migration to pay off, providers must develop a realistic migration strategy and goals, choose the appropriate cloud computing and services models, find one or more CSPs whose services, support, and pricing match their needs and ensure that their networks have the bandwidth capacity to handle cloud-based workloads.