In the increasingly competitive SaaS industry, being ‘always on’ has become non-negotiable. It means your application must be available, and performing optimally, 24/7/365. Any downtime or performance setbacks put your reputation, and bottom line, at risk.
In this blog we’ll explore what the concept of being always on really means for modern SaaS companies and why it’s so challenging to achieve without the right infrastructure in place.
Today’s SaaS platforms operate under a constant expectation of availability and customers won’t tolerate imperfection. SaaS applications need to function perfectly, all the time - and that means SaaS application hosting needs to be just as reliable.
As a cloud-native industry, SaaS organizations have traditionally relied on hyperscale cloud providers. Following the ‘cloud boom’, advertising hype surrounding hyperscale cloud’s elasticity, flexibility and minimal upfront costs made it the go-to choice. But, over time these environments have revealed additional complexities: spiraling bills, lack of hardware-level control, performance degradation, and over-reliance on single points of failure.
As SaaS market competition reaches new heights, hyperscale cloud-first strategies are fast becoming more of a competitive disadvantage. As a result, SaaS teams are landing on bare metal to achieve the reliability and performance they need.
Remaining always on is operationally and technically challenging as it involves remaining performant at high volume whilst also keeping margins tight and ensuring compliance. There’s no doubt that hyperscale cloud solutions excel in providing the rapid elasticity needed to scale fast. However, when relied on exclusively, these environments present additional challenges that will compromise a SaaS platform’s ability to keep the always on promise.
Hyperscale environments are built to provide massive elasticity, but it comes at a price. Virtualization overheads and unpredictable usage-based pricing means costs quickly spiral out of control. Research conducted by CloudZero found that for 73% of SaaS companies, cloud service provider costs represent at least 20% of their total cost of goods sold (and for the remaining 28% that ratio rises to more than half).
“For SaaS companies, the cloud is typically a top three budget line item - yet most don’t have rigorous processes for managing their cloud efficiency or understanding the relationship to their unit economics at any level of granularity,” comments Phil Pergola, CEO of CloudZero.
Pushing to remain always on whilst relying solely on hyperscale cloud contracts is an expense that most SaaS organizations won’t be able to maintain long-term.
On top of standard billing, data egress is another cost center. Public cloud providers charge egress fees each time data leaves their network. Since SaaS platforms handle vast volumes of data flowing across geographies, it means egress fees can become a significant expense.
Amazon S3, for instance, charges $0.02-0.09 per GB so syncing 10TB of data each month between us-east-1 and eu-west-1 regions could cost in excess of $900, depending on the regions involved. As a platform scales globally, egress fees spiral (even more so if replication is enabled and forgotten).
As Ievgen Nagorny, CTO of MGID, told servers.com:
“One of the prevailing concerns with cloud solutions is the unpredictability and non-transparency of costs. The fact that there are numerous companies providing cloud cost optimization services is testament to this problem. Businesses often find themselves paying for hidden costs or unnecessary resources, making budget management challenging”.
SaaS companies fall under stringent security and compliance requirements related to data protection and privacy. Even if your infrastructure runs entirely on virtualized infrastructure, you are ultimately responsible for securing your customer’s data, configuring your services compliantly, and adhering to required controls.
Ultimately, whilst a hyperscale cloud provider’s global infrastructure will allow SaaS platforms to operate in regions worldwide, it remains the responsibility of the SaaS platform to manage data residency compliance in accordance with local laws (GDPR, CCPA, UK/EU Data Acts).
Due to the limited manual control and transparency hyperscale cloud environments provide over underlying hardware and network layers, it can be difficult to monitor and uphold these provisions. For a multinational SaaS organization mobilizing these services, there’s always the risk of inadvertently storing customer data in non-compliant regions due to in-platform data misconfigurations that the business has little-to-no control or oversight over.
Part of being always on is remaining seamlessly operational even in the face of hardware failures and network outages. At the same time, one of the biggest misconceptions around hyperscale cloud is that it never fails - and it means that many businesses don’t have robust redundancy strategies in place.
The reality is that even the most established vendors or the most high-performance servers aren’t immune to failure. Even if a hyperscale cloud provider replicates its services across availability zones, vulnerabilities still exist.
Following the recent wave of outages hitting the likes of AWS and Cloudflare, it’s a fact that’s becoming harder to ignore. Isaac Douglas, CRO of servers.com explains:
“Cloudflare’s outage is a good reminder that ‘the cloud’ still runs on very real, very imperfect infrastructure. No matter how ‘premium’ the platform, it will have bad days: cloud providers go down, CDNs go down, DNS providers go down. The goal isn’t perfection; the goal is resilience.”
One thing is clear: in the race to stay always on no single vendor is an antidote to performance woes. Building resilience means avoiding single points of failure altogether.
Vendor outages aren’t the only risk to SaaS platform performance. Variable latency and slowed down response times are frustrating for end user customers and risk undermining service promises made to enterprise clients at sign up.
Loss of infrastructure control in virtualized, public cloud environments means that you’re dependent on the provider’s underlying network stack and that customization is limited. Plus, because these environments involve hardware sharing, there’s always the risk that another tenant’s heavy workload will degrade your own platform’s performance.
Add the notoriously slow and sparse support typically offered by hyperscalers into the mix and the chances of getting a quick fix to hardware issues compromising your platform’s performance are slim. For latency-sensitive SaaS platforms, this kind of multi-tenant, obfuscated environment introduces a level of performance variability that risks undermining the industry standard.
Continuous development, testing and integration is a core part of maintaining a high-performance SaaS platform. Testing must be completed at every stage of the SaaS platform development process to ensure regular bug fixes and updates. However, if not carefully managed, these testing workloads can increase the risk of temporary downtime from performance bottlenecks under the additional load (something that is exacerbated even further by the rapid release cycles typical of SaaS models).
In public cloud environments where there’s limited control over your hardware and less flexibility to define custom infrastructure tuning or hardware-specific testing, it becomes harder to maintain and monitor these testing environments efficiently. Software bugs and misconfigurations become harder to diagnose, noisy neighbors can result in overloaded servers, and the absence of robust failover mechanisms risks more serious downtime.
Influenced by these emerging challenges, longstanding attitudes to hyperscale cloud use are changing. SaaS platforms are looking beyond hyperscale cloud-first strategies and leaning on bare metal to mobilize several key benefits.
The first of these benefits is billing transparency. Bare metal solutions typically adopt flat-monthly or fixed pricing models as opposed to unpredictable, usage-based billing. Likewise, data transfer fees are often eliminated altogether or significantly lower than those charged for equivalent processes in hyperscale cloud environments.
It all contributes to greater billing transparency, making it easier to forecast infrastructure spend. For SaaS workloads that are relatively stable, this lowers overall unit cost. At the same time, it’s not unusual for this additional scope for optimization to result in further cost savings simply as a byproduct of effective resource utilization.
Simon Høiberg, CEO and founder of a SaaS portfolio that includes Aidbase, Linkdrio and Feedhive, is one of a growing community of SaaS leaders driving the shift to bare metal. In a recent social media post Høiberg shares how he successfully slashed his companies’ cloud bills from $7800 to $1000 by repatriating 80% of his SaaS portfolio to bare metal.
The second benefit is control. Dedicated hardware gives SaaS platforms full access to their underlying hardware resources without competition from ‘noisy neighbors’.
Without resource contention to contend with, it’s a win for reducing latency in SaaS. Bare metal environments allow for a far greater level of customization over hardware, networking and storage configurations. This means SaaS teams can create architectures optimized to suit the needs of their specific applications and improve overall performance and reliability.
Dedicated hardware also allows for a deeper level of control over security posture: you know exactly what hardware is in use, where data is stored, and who has access. This is a significant windfall for SaaS teams that need to be able to audit workflows to prove platform compliance. It also means that any continuous testing environments can be built with deeper tuning and customization. With more control over testing automation practices it’s far easier to consistently maintain good engineering hygiene.
The third benefit lies in support. In the race to stay always on the importance of exceptional support cannot be underestimated. Where hyperscalers offer break-fix solutions at most, a premium bare metal hosting provider will offer your team 24/7/365, tailored support. In the face of a hardware error or performance hiccup, having that continuous support is the only way to mitigate end-user impact quickly and effectively.
It's exactly why SaaS contact center solution, Voiso, chose bare metal from servers.com. CTO Gregor Potocar, shares: "[servers.com] support is responsive and knowledgeable, and they are always available to assist with any issues or questions that arise. Their proactive approach to customer service has made them a trusted partner in maintaining our infrastructure’s performance and reliability."
Put simply: if your SaaS platform has matured to the point where workloads are relatively predictable, performance is mission-critical, latency matters, and cost is under scrutiny, then bare-metal infrastructure becomes a strategic advantage. Not just a cost cutting move, but a route to reliability, performance and margin control.
Despite the challenges we’ve discussed, it’s not really a question of bare metal versus cloud. It’s about choosing the right infrastructure for the right workload. For many, the sweet spot lies in a hybrid SaaS approach leveraging hyperscale cloud elasticity and bare metal control.
In practice that typically means:
By placing workloads exactly where they fit best, this best of both approach ensures optimum performance is always maintained whilst eliminating any single point of failure.
The demand placed on modern SaaS companies is reshaping infrastructure decisions and leaders are recognising that bare metal offers a clear strategic advantage. While public cloud remains effective for providing elasticity, the presumption that placing all your infrastructure in hyperscale cloud will guarantee unfaltering performance has been debunked.
The reality for most SaaS organizations is that the best results lie in a hybrid approach, with bare metal forming the bedrock of core operations, and cloud supporting rapid elasticity.
If you’re running a SaaS platform and finding your cloud bills ballooning, latency compromising performance promises, or regulatory requirements becoming more stringent, now is the time to evaluate bare metal as part of your infrastructure strategy.
Unsure how to implement bare metal into your SaaS infrastructure? Reach out to our experts to learn more about our tailored bare metal cloud solutions and hybrid deployments.
Frances is proficient in taking complex information and turning it into engaging, digestible content that readers can enjoy. Whether it's a detailed report or a point-of-view piece, she loves using language to inform, entertain and provide value to readers.
![Reasons why companies are leaving the cloud [with examples]](/dA/845f390047/image/preview-why-companies-are-leaving-cloud.webp?language_id=1)