When - and why - hybrid SaaS might be a better option

Kurt Marko Profile picture for user kmarko October 7, 2020
A deep dive into alternatives to the multi-tenancy model used by most SaaS vendors.


Cloud services, particularly SaaS products like Office 365, G Suite, Slack and Zoom have been invaluable to enterprises trying to maintain functioning, productive operations amidst this year's sudden shift to remote work and distributed teams.

SaaS has long been popular as a replacement for internally operating broadly-used, commodified IT software like productivity suites and communications platforms.

However, the pandemic, which forced IT operations and support staff to work from home, led more organizations to investigate and adopt SaaS for backend business services like CRM, ERP and content management.

Indeed, Flexera's annual cloud survey found that one of the fastest-growing enterprise initiatives is migrating on-premises software to SaaS.

(Flexera 2020 State of the cloud report.)

The deeper SaaS creeps into an organization's back end services, the more it renders the availability and performance of critical business processes dependant upon a SaaS product and vendor that the business doesn't control. When revenue, customer satisfaction and business operations rely upon an external service provider, it forces business executives to fully understand the tradeoffs of SaaS and options for mitigating the risks without sacrificing the benefits. A hybrid SaaS implementation that combines shared and single-tenant infrastructure is often the right option. 

Shared- versus single-tenant SaaS

The SaaS deployment model is typically conflated with implementations using shared infrastructure. The concepts are understandably merged since the vast majority of SaaS products, and all consumer titles, are delivered from infrastructure shared across hundreds or thousands of customers. Nonetheless, the sensitive nature of customer, financial, procurement and HR data held in CRM, ERP, CMS and EHR systems led many software vendors to enter the SaaS market with single-tenant infrastructure to assuage the fears of skeptical business leaders. SAP S/4 HANA Cloud, STE is a prime example of the type that Brian Sommer warned about almost two years ago when he wrote:

Great SaaS apps run in the cloud in multi-tenancy (Old ERP vendors still peddle single-tenant on-premises, hosted and private cloud solutions).

Although Sommer lumps single-tenant SaaS together with on-premises software ("and their countless variants"), the single-tenant deployment model doesn't necessitate uncontrolled heterogeneity. Indeed, it is often built from a standardized software distribution, run on the same cloud environment as a vendor's multi-tenant customers, but using cloud resources, i.e. compute instances, storage columns and buckets and virtual networks, that are dedicated to one customer. Such implementations sacrifice some of the operational efficiencies of shared infrastructure for greater control over performance levels, maintenance windows and data placement.

A better SaaS alternative is a hybrid model in which parts of an application, such as the Web UI, mobile interface and administrative console are delivered from shared infrastructure while the backend databases and business logic run on dedicated infrastructure. For example, Generis CARA, a content management SaaS that targets highly regulated industries like life sciences and financial services provides two classes of service that vary the mix between shared and single-tenant resources: 

  • Standard with shared infrastructure for the application logic and UI servers, with a dedicated database for each customer. Updates are automatically applied each quarter.
  • VIP with dedicated services for the application middleware, UI and database. Updates are available each quarter, with customer control over the schedule (including the option of skipping updates)

The advantages of a hybrid approach include:

  • Predictable application performance by eliminating resource contention.
  • Customer control over software updates and maintenance downtime.
  • Customer control over application version and configuration, backup policies and the physical location of data repositories. 
  • Lower lock-in risk by using dedicated databases that are more easily replicated or migrated to other cloud infrastructures. 
  • Faster data restoration in the event of accidental or malicious deletion, again due to not comingling customers on a shared database or storage repository.

Noisy neighbors and other cloud performance concerns

Data security and cost control usually top the list of business concerns when asking executives about cloud usage, however, dig deeper and you'll find that many IT leaders also worry about performance when using SaaS applications. An ESG survey focusing on applications for distributed workers, i.e. every employee in the COVID era, found that dismal SaaS performance for remote workers is common.

(Data from ESG report; The Impact of Poor SaaS Performance on GloballyDistributed Enterprises. Chart from the author. )

Furthermore, 29% of respondents say that SaaS slowdowns have a "significant impact" on their business while another 61% say slow performance has some effect. Therefore, it's not surprising that virtually all ESG's respondents, an overwhelming 96%, say a solution that accelerated SaaS performance would be valuable. Indeed, the share saying such an option would be "very valuable" was 10 points higher (55%) in organizations where most of their employees were remote, i.e. resembled the post-pandemic norm in all organizations.

My take

SaaS introduces several potential bottlenecks that can degrade application performance and availability. As the ESG report intimates, inadequate network bandwidth and high latency is a frequent culprit for remote users. Networking problems have been exacerbated this year when an employee might be sharing a low-bandwidth broadband connection with a working spouse and two children still taking classes by video. However, as employees adjust to a quasi-permanent WFH environment, most can accommodate network problems via higher-speed broadband circuits and SD-WAN acceleration. Future 5G fixed wireless service and increased fiber-to-the-home buildout offer a long-term fix via gigabit-per-second service.

Another significant source of SaaS problems is resource contention and maintenance operations on shared infrastructure, which can be solved via single-tenant or hybrid SaaS designs. Most SaaS vendors rightly focus on shared infrastructure to maximize efficiency and offer compelling price points, however, that might change as more offer alternative user- or value-based pricing models. Indeed, data shows that 40% of SaaS companies now set prices based on value metrics where customers are willing to pay more for better performance, availability, application functionality and KPI improvements. 

SaaS growth projections of 15% annual increases in revenue over the next few years are already outdated as companies adjusted to pandemic lockdowns and remote work by flocking to cloud services. Indeed, Salesforce has seen sales grow by almost 30% over the last year, while Adobe's revenue increased by 24%. As SaaS becomes more broadly used and further ingrained in critical business processes, expect to see new deployment and pricing models that reflect the unmet needs of particular industries and use cases. A hybrid tenancy SaaS deployment that uses dedicated infrastructure for some subsystems will be attractive to organizations hesitant to migrate familiar, predictable on-premises software to SaaS.


Disclosure - I occasionally consult with and do content development for technology companies, including Generis Corporation. However, it had no role in or editorial authority over this column. At time of writing, Salesforce is a premier partner of diginomica.

A grey colored placeholder image