Jason Aw
Healthcare information systems, high availability, and the cloud
August 28, 2020
By Jason Aw
Information is ultimately the lifeblood of any healthcare system. You know the forms it takes: Patient records, standard operating procedures, claims, PPE inventory levels. Yes, there are costly MRIs and surgical suites. There are frontline doctors and nurses (and even more people behind the front line), but even they are tethered to information: Their time needs to be scheduled and tracked; they need information about the next person they’re going to see; they add more information to the EHR. It’s all bits and bytes; it all needs to be captured, organized, and presented to the person who needs to perform the next action—whether that’s to save a life, process a claim, order supplies, or prepare a room for the next patient.
To say the least, it would be utter pandemonium if suddenly the information everyone needed were unavailable.
Healthcare information systems consultants and healthcare information technology (IT) providers are well aware of the importance of information availability. Back in the day, when healthcare IT systems were built on-premises, the IT teams configured physically redundant systems, with one system standing on standby in a remote location. If an active system crashed, mechanisms were in place to fail over to the standby system, which could immediately provide continued access to critical information.
Today, though the cloud is becoming an increasingly attractive home for the deployment of healthcare technology. MEDITECH, to take but one example, is offering its entire Expanse electronic health record (EHR) system as a web-based system that lives in the cloud. For MEDITECH and its customers, offering its EHR system as a cloud-based solution can accelerate initial deployment, facilitate feature development, and optimize costs.
But healthcare IT departments need to take special precautions when deploying any critical IT solutions that will live in the cloud. The major cloud infrastructure providers—AWS, Azure, and Google—provide impressive infrastructure availability and durability guarantees, but those guarantees really are for infrastructure only. They can guarantee that at least one virtual machine (VM) in your infrastructure will be accessible and operating 99.99% of the time. That translates to less than an hour of downtime over the course of a year. If your critical EHR system is running on that infrastructure, that may sound like a level of vulnerability that you can live with—except that the infrastructure guarantee doesn’t guarantee the availability of your critical EHR system.
You can configure your critical healthcare IT solutions for true high availability in the cloud—and gain all the advantages that the cloud affords in terms of ease of access, rapid roll-out, and lower on-site infrastructure costs. But you need to configure for high availability with application access in mind, not just hardware access.
When it comes to high availability application access, one of the key challenges lies in how and where you store the data captured by your healthcare IT systems. You can’t rely on a traditional storage area network (SAN) the way you might have in an on-premises failover configuration because you can’t share storage in the cloud. There are a wide range of storage options that you can deploy—from software-defined storage offerings by third parties to high speed SSD-based storage offerings offered by the cloud service providers themselves. Whatever you choose, you’re going to need to attach storage to both your primary and backup servers, which you’ll configure as a failover cluster, with the primary and secondary systems residing in different data centers (or different Availability Zones if you’re deploying on AWS). You’ll also need a mechanism for actively synchronizing the data from the primary to the secondary storage systems so that, should the primary system fail, your secondary system has an up-to-date copy of all the data that the primary system had been using and can immediately stand in for the primary system that has gone offline.
There are a number of ways to manage replication between storage instances, but you’ll need to test them to be confident that they support your availability targets. The Chris O'Brien Lifehouse hospital in Sydney, Australia, initially configured their AWS-based MEDITECH deployment for high availability using a software-defined storage volumes they selected through the AWS marketplace. The replication and failover features of the storage solution performed well when tested. The team was confident that the backup infrastructure would be online and supporting teams within moments. But the software-defined storage volumes presented an unexpected problem: the EHR system itself delivered sub-par performance when interacting with the software-defined storage. The IT team at Chris O’Brien Lifehouse reconfigured the underlying infrastructure—which they could do rapidly in the cloud—and attached a native AWS storage solution to each failover cluster node, and the EHR solution immediately began to deliver the throughput the team was expecting.
At that point, the IT team at Chris O’Brien Lifehouse worked with the team at AWS to deploy a third-party SANless clustering tool to enable high-performance data replication among the storage systems in the failover cluster, which then ensured that the secondary server nodes all had up-to-date copies of all EHR data. If for any reason the primary infrastructure failed, the secondary infrastructure could come online in moments and enable continued access to the information that individuals and systems throughout the Chris O’Brien Lifehouse needed.
In the end, the cloud offers healthcare IT providers—and healthcare information consumers—a great deal of flexibility and power. You can configure for the high availability that healthcare requires, but you need to think differently about how you will achieve your availability goals and make sure you test all aspects of your cloud configuration to ensure that your solution achieving one goal is not compromising your solution for achieving another.
About the Author: Jason Aw, technical lead, APAC at SIOS Technology, is an IT professional who has been focused on high availability and disaster recovery for over 20 years. He has helped hundreds of enterprises implement HA/DR solution to protect critical applications including SQL, Oracle databases, SAP, web and many others.
Information is ultimately the lifeblood of any healthcare system. You know the forms it takes: Patient records, standard operating procedures, claims, PPE inventory levels. Yes, there are costly MRIs and surgical suites. There are frontline doctors and nurses (and even more people behind the front line), but even they are tethered to information: Their time needs to be scheduled and tracked; they need information about the next person they’re going to see; they add more information to the EHR. It’s all bits and bytes; it all needs to be captured, organized, and presented to the person who needs to perform the next action—whether that’s to save a life, process a claim, order supplies, or prepare a room for the next patient.
To say the least, it would be utter pandemonium if suddenly the information everyone needed were unavailable.
Healthcare information systems consultants and healthcare information technology (IT) providers are well aware of the importance of information availability. Back in the day, when healthcare IT systems were built on-premises, the IT teams configured physically redundant systems, with one system standing on standby in a remote location. If an active system crashed, mechanisms were in place to fail over to the standby system, which could immediately provide continued access to critical information.
Today, though the cloud is becoming an increasingly attractive home for the deployment of healthcare technology. MEDITECH, to take but one example, is offering its entire Expanse electronic health record (EHR) system as a web-based system that lives in the cloud. For MEDITECH and its customers, offering its EHR system as a cloud-based solution can accelerate initial deployment, facilitate feature development, and optimize costs.
But healthcare IT departments need to take special precautions when deploying any critical IT solutions that will live in the cloud. The major cloud infrastructure providers—AWS, Azure, and Google—provide impressive infrastructure availability and durability guarantees, but those guarantees really are for infrastructure only. They can guarantee that at least one virtual machine (VM) in your infrastructure will be accessible and operating 99.99% of the time. That translates to less than an hour of downtime over the course of a year. If your critical EHR system is running on that infrastructure, that may sound like a level of vulnerability that you can live with—except that the infrastructure guarantee doesn’t guarantee the availability of your critical EHR system.
You can configure your critical healthcare IT solutions for true high availability in the cloud—and gain all the advantages that the cloud affords in terms of ease of access, rapid roll-out, and lower on-site infrastructure costs. But you need to configure for high availability with application access in mind, not just hardware access.
When it comes to high availability application access, one of the key challenges lies in how and where you store the data captured by your healthcare IT systems. You can’t rely on a traditional storage area network (SAN) the way you might have in an on-premises failover configuration because you can’t share storage in the cloud. There are a wide range of storage options that you can deploy—from software-defined storage offerings by third parties to high speed SSD-based storage offerings offered by the cloud service providers themselves. Whatever you choose, you’re going to need to attach storage to both your primary and backup servers, which you’ll configure as a failover cluster, with the primary and secondary systems residing in different data centers (or different Availability Zones if you’re deploying on AWS). You’ll also need a mechanism for actively synchronizing the data from the primary to the secondary storage systems so that, should the primary system fail, your secondary system has an up-to-date copy of all the data that the primary system had been using and can immediately stand in for the primary system that has gone offline.
There are a number of ways to manage replication between storage instances, but you’ll need to test them to be confident that they support your availability targets. The Chris O'Brien Lifehouse hospital in Sydney, Australia, initially configured their AWS-based MEDITECH deployment for high availability using a software-defined storage volumes they selected through the AWS marketplace. The replication and failover features of the storage solution performed well when tested. The team was confident that the backup infrastructure would be online and supporting teams within moments. But the software-defined storage volumes presented an unexpected problem: the EHR system itself delivered sub-par performance when interacting with the software-defined storage. The IT team at Chris O’Brien Lifehouse reconfigured the underlying infrastructure—which they could do rapidly in the cloud—and attached a native AWS storage solution to each failover cluster node, and the EHR solution immediately began to deliver the throughput the team was expecting.
At that point, the IT team at Chris O’Brien Lifehouse worked with the team at AWS to deploy a third-party SANless clustering tool to enable high-performance data replication among the storage systems in the failover cluster, which then ensured that the secondary server nodes all had up-to-date copies of all EHR data. If for any reason the primary infrastructure failed, the secondary infrastructure could come online in moments and enable continued access to the information that individuals and systems throughout the Chris O’Brien Lifehouse needed.
In the end, the cloud offers healthcare IT providers—and healthcare information consumers—a great deal of flexibility and power. You can configure for the high availability that healthcare requires, but you need to think differently about how you will achieve your availability goals and make sure you test all aspects of your cloud configuration to ensure that your solution achieving one goal is not compromising your solution for achieving another.
About the Author: Jason Aw, technical lead, APAC at SIOS Technology, is an IT professional who has been focused on high availability and disaster recovery for over 20 years. He has helped hundreds of enterprises implement HA/DR solution to protect critical applications including SQL, Oracle databases, SAP, web and many others.